Monday, December 15, 2008

IT Systems Analytics Become Crucial as Move to Cloud and SaaS Raises Complexity Bar

Transcript of a BriefingsDirect podcast on the role of log management and analytics as enterprises move to cloud computing and software as a service.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Learn more. More related podcasts. Sponsor: LogLogic.

Dana Gardner: Hi, this is Dana Gardner, principal analyst at Interarbor Solutions, and you're listening to BriefingsDirect. Today, we present a sponsored podcast discussion on the changing nature of IT systems' performance and the heightening expectations for applications delivery from those accessing application as services.

The requirements and expectations on software-as-a-service (SaaS) providers are often higher than for applications traditionally delivered by enterprises for their employees and customers. Always knowing what's going on under the IT hood, being proactive in detection, security, and remediation, and keeping an absolute adherence to service level agreements (SLAs), are the tougher standards a SaaS provider deals with.

Increasingly, this expected level of visibility, management, and performance will apply to those serving up applications as services regardless of their hosting origins or models.

Here to provide the full story on how SaaS is making all applications' performance expectations higher, and how to meet or exceed those expectations is Jian Zhen, senior director of product management at LogLogic. Welcome to the show Jian.

Jian Zhen: Thank you for having me.

Gardner: We're also joined by Phil Wainewright, an independent analyst, director of Procullux Ventures, and SaaS blogger at ZDNet and ebizQ. Welcome back to the show, Phil.

Phil Wainewright: Glad to be here, Dana.

Gardner: Phil, let’s start with you. The state of affairs in IT is shifting. Services are becoming available from a variety of different models and hosts. We're certainly hearing a lot about cloud and private cloud. I suppose the first part of this that caught the public's attention was this whole SaaS notion and some successes in the field for that.

Maybe you could help us understand how the world has changed around SaaS infrastructure, and what implications that has for the IT department?

Wainewright: One thing that's happening is that the SaaS infrastructure is getting more complicated, because more choice is emerging. In the past people might have gone to one or two SaaS vendors in very isolated environments or isolated use cases. What we're now finding is that people are aggregating different SaaS services.

They're maybe using cloud resources alongside of SaaS. We're actually looking at different layers of not just SaaS, but also platform as a service (PaaS), which are customizable applications, rather than the more packaged applications that we saw in the first generation of SaaS. We're seeing more utility and cloud platforms and a whole range of options in between.

That means people are really using different resources and having to keep tabs on all those different resources. Where in the past, all of an IT organizations' resources were under their own control, they now have to operate in this more open environment, where trust and visibility as to what's going on are major factors.

Gardner: Do you think that the type of application delivery that folks are getting from the Web will start to become more the norm in terms of what delivery mechanisms they encounter inside the firewall from their own data center or architecture?

Wainewright: If you're going to take advantage of SaaS properly, then you need to move to more of a service-oriented architecture (SOA) internally. That makes it easier to start to aggregate or integrate these different mashups, these different services. At the end of the day, the end users aren't going to be bothered whether the application is delivered from the enhanced data center or from a third-party provider outside the firewall, as long as it works and gives them the business results they're looking for.

Gardner: Let's go to Jian Zhen at LogLogic. How does this changing landscape in IT and in services delivery affect those who are responsible for keeping the servers running, both from the host as well as the receiving end in the network, and those who are renting or leasing those applications as services?

Zhen: Phil hit the nail on the head earlier when he mentioned that IT not only has to keep track of resources within their own environment, but now has to worry about all these resources and applications outside of their environment that they may or may not have control over.

That really is one of the fundamental changes and key issues for current IT organizations. You have to worry not only about who is accessing the information within your company firewall, but now you have all this data that's sitting outside of the firewall in another environment. That could be a PaaS, as Phil said, it could be a SaaS, an application that's sitting out there. How do you control that access? How do you monitor that access. That's one of the key issues that IT has to worry about.

Obviously, there are data governance issues and activity monitoring issues. Now, from a performance and operational perspective, you have to worry about, are my systems performing, are these applications that I am renting, or platforms or utilities I am renting, are they performing to my spec? How do I ensure that the service providers can give me the SLAs that I need.

Those are some of the key issues that IT has to face when they are going outside of this corporate firewall.

Gardner: I suppose if it were just one application that you knew you were getting as a service, if something would go wrong, you might have a pretty good sense of who is responsible and where, but we are very rapidly advancing toward mixtures, hybrids, multiple SaaS providers, different services that come together to form processes. Some of these might be on premises, and some of them might not be.

It strikes me that we're entering a time when finger pointing might become rampant if something goes wrong, who is ultimately responsible, and under whose SLA does it fall?

Phil, from your perspective, how important will it be to gain risk, compliance, and security comfort, by being able to quickly identify who is the source of any issue?

Wainewright: That's vitally important, and this is a new responsibility for IT. To be honest Dana, you're a little bit generous to the SaaS providers when you say that if you only dealt with one or two, and if something went down, you had a fair idea of what was going on. What SaaS providers have been learning is that they need to get better at giving more information to their customers about what is going wrong when the service is not up or the service is not performing as expected. The SaaS industry is still learning about that. So, there is that element on that side.

On the IT side, the IT people have spent too much time worrying about reasons why they didn't want to deal with SaaS or cloud providers. They've been dealing with issues like what if does go down, or how can I trust the security? Yes, it does go down sometimes, but it's up 99.7 percent of the time or 99.9 percent of the time, which is better than most organizations can afford to do with their own services.

Let's shift the emphasis from, "It's broken, so I won't use it," to a more mature attitude, which says, "It will be up most of the time, but when it does break, how do I make sure that I remain accountable, as the IT manager, the IT Director, or the CIO. How do I remain accountable for those services to my organization, and how do I make sure that I can pinpoint the cause of the problem, and get it rectified as quickly as possible?"

Gardner: Jian, this offers a pretty significant opportunity, if you, as a vendor and a provider of services and solutions, can bring visibility and help quickly decide where the blame lies, but I suppose more importantly, where the remediation lies. How do you view that opportunity, and what specifically is LogLogic doing?

Zhen: We talked to a lot of customers who were either considering or actually going into the cloud or using SaaS applications. One of the great quotes that we recently got from a customer is, "You can outsource responsibility, but not accountability." So, it fits right into what Phil what was saying about being accountable and about your own environment.

The requirement to comply with government regulations and industry mandates really doesn't change all that much, just because of SaaS or because a company is going into the cloud. What it means is that the end users are still responsible for complying with Sarbanes-Oxley (SOX), payment cared industry (PCI) standards, the Health Insurance Portability and Accountability Act (HIPAA), and other regulations. It also means that these customers will also expect the same type of reports that they get out of their own systems.

IT organizations are used to transparency in their own environment. If they want to know what's happening in their own environment, they can get access to it. They can at least figure out what's going on. As you go into the cloud and use some of the SaaS applications, you start to lose some of that transparency, as you move up the stack. Phil mentioned earlier, there's infrastructure as a service, PaaS, SaaS. As you go up the stack, you're going to lose more and more of that transparency.

From a service-provider perspective, we need these providers to provide more transparency and more information as to what's happening in their environment and who has access. Who did access the information? LogLogic's can help these service providers get that kind of information and potentially even provide the reports for their end users.

From a user's perspective, there is that expectation. They want to know what's going on and who is accessing the data. So, the service providers need to have the proper controls and processes in place, and need to continuously monitor their own infrastructure, and then provide some of these additional reports and information to their end customers as needed.

Gardner: LogLogic is in the business of collating and standardizing information from a vast array of different systems through the log files and other information and then offering reports and audit capabilities from that data. It strikes me that you are now getting closer to what some people call business intelligence (BI) for IT, in that you need to deal almost in real time with vast amounts of data, and that you might need to adjust across boundaries in order to gain the insights and inference.

Do you at LogLogic cotton to this notion of BI for IT, and if so, what might we expect in the future from that?

Zhen: BI for IT or IT intelligence, as I have used the term before, is really about getting more information out of the IT infrastructure; whether it's internal IT infrastructure or external IT infrastructure, such as the cloud.

Traditionally, administrators have always used logs as one of the tools to help them analyze and understand the infrastructure, both from a security and operational perspective. For example, one of the recent reports from Price Waterhouse, I believe, says that the number one method for identifying security incidents and operational problems is through logs.

LogLogic's can provide the infrastructure and the tools to help customers gather the information and correlate different log sources. We can provide them that information, both from an internal and external perspective. We work with a lot of service providers, as you know, companies like SAVVIS, VeriSign, Verizon Business Services, to provide the tools for them to analyze service provider infrastructures as well.

A lot of that information can be gathered into a central location, correlated, and presented as business intelligence or business activity monitoring for the IT infrastructure.

Gardner: Phil, the amount of data that we can extract from these systems inside the service providers is vast. I suppose what people are looking for is the needle in the haystack. Also, as you mentioned, it probably behooves these providers to offer more insights into how well they did or didn't do.

What's your take on this notion of BI for IT, and does it offer the SaaS providers an opportunity to get a higher level of insight and detail about what is going on within their systems for the assurance and risk mediation for their customers?

Wainewright: Yes, it does. This is an area where we are going to see best practices emerge. We're in a very early stage. Talking about keeping logs reminds me of what happened in the early days of Web sites and Web analytics. When people started having Web sites, they used to create these log files, in which they accumulated all this data about the traffic coming to the site. Increasingly, it became more difficult to analyze that traffic and to get the pertinent information out.

Eventually, we saw the rise of specialist Web-traffic analytics vendors, most of them, incidentally, providing their services as SaaS focused on helping the Web-site managers understand what was going on with their traffic.

IT is going to have to do the same thing. Anyone can create a log file, dump all the data into a log, and say that they've got a record of what's been going on. But, that's the technically easy challenge. The difficult thing, as Jian said, is actually doing the business analytics and the BI to see what was going on, and to see what the information is.

Increasingly, it comes back to IT accountability. If your service provider does go down, and if the logs show that the performance was degrading gradually over a period of time, then you should have known that. You should have been doing the analysis over time, so that you were ahead of that curve and were able to challenge the provider before the system went down.

If it's a good provider, which comes back to the question you asked, then the provider should be on top of that before the customer finds out. Increasingly, we'll see the quality of reporting that providers are doing to customers go up dramatically. The best providers will understand that the more visibility and transparency they provide the customers about the quality of service they are delivering, the more confidence and trust their customers will have in that service.

Gardner: As we mentioned, the expectations are increasing. The folks who rent an application for a few dollars a month actually have higher expectations on performance than perhaps far more expensive applications inside a firewall and the traditional delivery mechanisms.

Wainewright: That's right, Dana. People get annoyed when Gmail goes down, and that's free. People do have these high expectations.

Gardner: Perhaps we can meet those expectations, even as they increase, but even more importantly for these providers is the cost at which they deliver their services. The utilization rates, the amount of energy that’s required per task or some metric like that, these log files, and this BI will decide their margins and how competitive they are in what we expect to be a fairly competitive field. In fact, we are starting to see the signs of marketplace and auctioning types of activities around who can put up a service for the least amount of money, which, of course, will put more downward pressure on margin.

I've got to go back to Jian on this one. We can certainly provide for user expectations and SLAs, but ultimately how well you run your data center as a service provider dictates your survival ability or viability as a business.

Zhen: You're absolutely right. One of the things that service providers, SaaS providers, or cloud providers have always talked about is the economy of scale. Essentially, that's doing more with less in order to understand your IT infrastructure and understand your customer base. This is what BI is all about, right? You're analyzing your business, your user base, the user access, and all that information in trying to come up with some competitive advantage to either reduce cost or increase efficiency.

All that information is in logs, whether logs that are spewed out by your IT infrastructure, logs that are instrumented using agents or application performance, monitoring type of tools. That information is there, and you need to be able to automate and enhance the ways things are done. So, you need to understand and see what's going on in the environment.

Analyzing all those logs gives you critical capability, not only managing hundreds or thousands of systems and making them more efficient, but bringing that BI throughout. Seeing how your users are accessing, reacting to, or changing your system makes it more efficient for the user, faster for the user, and, at the same time, reduces that cost to manage the infrastructure, as well as to do business.

So, the need to understand and see what's going on is really driving the need to have better tools to do system analysis.

Gardner: Well, how about that Phil? With apologies to Monty Python, every electron is important, right?

Wainewright: Well, it certainly can be. I think the other benefits of providers monitoring this information is that, if they can build out a track record and demonstrate that they all providing better service, then maybe that's the way of defending themselves, of being able to justify asking higher prices than they might otherwise have done.

If the pricing is going to go down because of competitive pressures, there will be differential pricing according to the quality that providers can show they have a track record for delivering.

Zhen: I definitely agree with that. Being able to provide better SLAs, being able to provide more transparency, audit transparency, are things that enterprises care about. As many reports have mentioned, it's one of the biggest issues that's preventing enterprises from adopting the cloud or some of these SaaS applications. Not that the enterprises are not adopting, but the movement is still very slow.

The main reasons are security and transparency. As SaaS providers or service providers start providing a lot more information based on the data that they analyze, they can provide better SLAs, both from an uptime and performance perspective, not just uptime. A lot of the SLAs today just talk about uptime. If they can provide a lot of that information by analyzing the information that they already have -- the log data, access data, and what not -- that’s a competitive advantage for the providers. They can charge a higher price, and often, enterprises are willing to pay for that.

Wainewright: I've been speaking to enterprise customers, and they are looking for better information from the providers about those performance metrics, because they want to know what the quality of service is. They want to know that they're getting value for money.

Gardner: Well, we seem to have quite a set of pressures. One, to uphold performance, provide visibility, reduce risk, and offer compliance and auditing benefits. On the other side, it's pure economics. The more insight and utilization you have, and the more efficiently you can run your data centers, the more you can increase your margin and scale out to offer yet more services to more types of customers. It seems pretty clear that there's a problem set and a solution set.

Jian, you mentioned that you had several large service providers as customers. I don’t suppose they want all the details about what happens inside their organizations to come out, but perhaps you have some use case scenarios. Do you have examples of how analytics from a system’s performance, vis-à-vis log data, helps them on either score, either qualitatively in terms of performance and trust, and more importantly, over time, their ability to reap the most efficiency out of their system?

Zhen: These are actually partners of LogLogic. We've worked with these service-provider partners to provide managed services or cloud services for log management to the end customers. They're using it both working with the customers themselves, as well as using it internally.

Often, the use cases are really around compliance and security. That’s where the budget is coming from. Compliance is the biggest driver for some of these tools today.

However, some of the reports I mentioned, especially from Enterprise Strategy Group (ESG), one of the fastest-growing use cases for log management is operational use. This means troubleshooting, forensic analysis, and being able to analyze what's going on in the environment. But, the biggest driver today for purchasing that type of log-management solution is still compliance -- being able to comply with SOX, PCI, HIPAA, and other regulations.

Gardner: Let’s wrap up with some crystal-ball gazing. First, from Phil. How do you see this market shaking out? I know we're under more economic pressure these days, given the pending or imminent global recession, but it seems to me that it could be a transformative pressure, a catalyst, toward more adoption of services, and keeping application performance at lowest possible cost. What's your sense of where the market is going.

Wainewright: It’s a terrible cliché, but it’s about doing more with less. It may be a cliché, but it’s what people are trying to do. They've got to cut costs as organizations, and, at the same time, they have to actually be more agile, more flexible, and more competitive.

That means a lot of IT organizations are looking to SaaS and they're looking to cloud computing, because this is the way of getting resources without a massive outlay and starting to do things with a relatively low risk of failure.

They're finding that budgets are tight. They need to get things done quickly. Cloud or SaaS allows them to do that, and therefore there's a rosy future, even in bleak economic conditions, for this type of offering.

There are still a lot of worries among IT people as to the reliability and security and privacy compliance and all the other factors around SaaS. Therefore, the SaaS providers have to make sure that they're monitoring that, and that they're reporting. Likewise, the IT people, for their own peace of mind, need to make their own arrangement, so that they can also be keeping an eye on their side. I think everyone is going to be tracking and monitoring each other.

The upside of is that we're going to get more enterprise-class performance and enterprise-class infrastructure being built around the cloud services and the SaaS providers, so that enterprises will be able to have more confidence. So, at the end of the economic cycle, once people start investing again, I think we'll see people continue to invest in cloud services and SaaS, not because it's the low-cost option, but because it's the proven option that they have confidence in.

Gardner: Jian Zhen, how do you and LogLogic see the market unfolding? Where do you think the opportunities lie?

Zhen: I definitely agree with Phil. With the current economic environment, a lot of enterprises will start looking at SaaS and cloud services seriously and consider them.

However, enterprises are still required to be compliant with government regulations and industry mandate, so that's not going to go away. For the service providers and the SaaS providers, what they can do to attract these customers really is to make themselves more attractive, and make themselves be compliant with some of these regulations, and provide more transparency, giving people a view into who is accessing the data, and how they protect the data.

Amazon did a great thing, which was to release a white paper on some of their security practices. It's a very high level, but it’s a good start. Service providers need to start thinking more along the lines of, how to attract these enterprise customers, because the enterprise customers are willing and seriously considering SaaS services.

Phil had an article a while back, calling for a SaaS code of conduct. Phil, one of the things that you should definitely add there is a code to have the service providers provide all the transparency. That’s a thing that service providers can use to offer essentially a competitive advantage for their enterprise customers.

Gardner: Now, you sit at a fairly advantageous point, or a catbird's seat, if you will, on this regulatory issue. As enterprises seek more SaaS and cloud services for economic and perhaps longer-term strategic reasons, do we need to rethink some of our compliance and regulatory approaches?

We have a transition in the United States in terms of the government. So, now is a good time, I suppose, to look at those sorts of things. What, from your perspective, should change in order to allow companies to more freely embrace and use cloud and SaaS services, when it comes to regulation and compliance?

Zhen: As far as changing the regulations, I'm not sure there are a lot of things. We've seen SOX become a very high level and very costly regulation to be compliant with. However, we've also have seen PCI. That’s much more specific, and companies and even service providers can adopt and use some of these requirements.

Gardner: That's the payment card issue, right?

Zhen: Correct. The PCI data-security standard is a lot more specific as to what a company has to do in order to be compliant with it. Actually, one of the appendixes is really for service providers. A lot of service providers have used, for example, the Statement on Auditing Standards (SAS) 70 Type II kind of a report as one of the things they show the customer that they are compliant with. However, I don’t think the SAS 70 Type II is sufficient, mainly because the controls are described by the service providers themselves.

Essentially, they set their own requirements and they say, "Hey, we meet these requirements." I don’t think that’s sufficient. It needs to be something that’s more industry standard, like PCI, but maybe a little bit different, definitely more specific as to what the service providers needs to do.

On top of that, we need some kind of information on when security incidents happen with service providers. One of the things that 44 states have today is data-breach notification laws. That law obviously doesn’t apply to SaaS providers, but in order to provide more transparency there may need to be some standard or some processes in how breaches are reported and handled.

Some of these things certainly will help enterprises be more comfortable in adopting the services.

Gardner: Well, there are some topics Phil for about 150 blog entries, this whole notion of how to shift regulation and compliance in order to suit a cloud economy.

Wainewright: Yeah, it's going to be a difficult issue for the cloud providers to adapt to, but a very important one. This whole issue of SAS 70 Type II compliance, for example. If you're relying on a service provider for part of the services that you provide, then your SAS 70 Type II needs to dovetail with their SAS 70 Type II processes.

That’s the kind of issue that Jian was alluding to. It's no good just having SAS 70 Type II, if the processes that you've got are somehow in conflict with or don't work in collaboration with the service providers that you are depending on. We have to get a lot smarter within the industry about how we coordinate services and provide accountability and audit visibility and trackability between the different service providers.

Gardner: Very good. We've been discussing requirements and expectations around SaaS providers, looking at expected increases and demands for visibility, and management and performance metrics. Helping us to better understand these topics -- and I'm very happy that they joined us -- are Jian Zhen, senior director of product management at LogLogic. Thanks for your input, Jian.

Zhen: Thank you, Dana.

Gardner: Also Phil Wainewright, independent analyst, director of Procullux Ventures, and SaaS blogger at ZDNet and ebizQ. Always good to have you here Phil, thank you.

Wainewright: Thanks, Dana.

Gardner: This is Dana Gardner, principal analyst at Interarbor Solutions. You've have been listening to a sponsored BriefingsDirect podcast. Thanks, and come back next time.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Learn more. More related podcasts. Sponsor: LogLogic.

Transcript of a BriefingsDirect podcast on the role of log management and analytics as enterprises move to cloud computing and SaaS. Copyright Interarbor Solutions, LLC, 2005-2008. All rights reserved.

Sunday, December 14, 2008

BriefingsDirect Analysts Handicap Large IT Vendors on How Cloud Trend Impacts Them

Edited transcript of BriefingsDirect Analyst Insights Edition podcast, Vol. 34, on cloud computing and its impact on IT vendors, recorded Nov. 21, 2008.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Charter Sponsor: Active Endpoints.

Special offer: Download a free, supported 30-day trial of Active Endpoint's ActiveVOS at

Dana Gardner: Hello, and welcome to the latest BriefingsDirect Analyst Insights Edition, Vol. 34. This periodic discussion and dissection of IT infrastructure related news and events with a panel of industry analysts and guests comes to you with the help of our charter sponsor, Active Endpoints, maker of the ActiveVOS, visual orchestration system. I'm your host and moderator, Dana Gardner, principal analyst at Interarbor Solutions.

Our topic this week, the week of Nov. 17, 2008 -- and Happy Holidays to you all -- is the gathering cloud-computing conundrum. From the hype and intrigue that's been a top story of 2008, do you think that cloud computing is past its infancy and well into adolescence?

We really are only beginning to understand how the IT services delivery, data management, and economic models of cloud computing will impact the market. Today, we're going to focus on the impact that this coalescing cloud phenomenon and the myriad cloud-computing definitions will have on the large, established IT vendors.

If this shift is as large and inevitable as many of us think, the impact on the current IT business landscape will also be large. Some will do well, and some will not. All, I expect, will need to adapt, and the shifts are certainly exacerbated by the deepening global recession.

To help us dig into how cloud computing will impact the IT industry we're joined by this week's panel. I'd like to welcome Jim Kobielus, senior analyst at Forrester Research. Hey, Jim.

Jim Kobielus: Hi, Dana. Hi, everybody.

Gardner: Tony Baer, senior analyst at Ovum. Hello, Tony.

Tony Baer: Hey, Dana, good to speak to you again.

Gardner: We're also joined by Brad Shimmin, principal analyst at Current Analysis. Welcome, Brad.

Brad Shimmin: Thanks for having me.

Gardner: Joe McKendrick, independent analyst and prolific blogger on ZDNet and ebizQ. Hi Joe.

Joe McKendrick: Hey, Dana, great to be here.

Gardner: This cloud chatter just doesn't seem to want to go away. In some respects, it's being heightened by the recession and this added emphasis on doing more with less, cutting down on IT spend and operation costs, the whole pressure on reducing the burden, if you will, that IT can provide for companies, even as they might be contracting.

We're also mixing in virtualization, services-oriented architecture (SOA), governance, platform as a service (PaaS), software as a service (SaaS), and social networking into this cloud-services definition mix.

Many see this cloud as joining, or at least somewhat mashing up or overlapping, the business-to-business (B2B) world with the business-to-consumer (B2C) world. In other words, IT provides services inside of organizations but now the Internet can provide some, and maybe many, of those services, regardless of whether they are a company or individual. So, there's a bit of competition between internal and external IT services.

Let me go to you first, Tony Baer. Of all the large IT vendors -- and you've been covering these folks for a long time -- what's your knee-jerk reaction? Which one has the best outlook from these cloud trends, and which one has the worst?

Baer: That's an interesting question. I hadn't really thought about it in that manner, but I'll just think out loud. Obviously, if we are going to talk about who has consistently positioned themselves as being the poster child, it has been Marc Benioff over at, where they have evolved from a customer relationship management (CRM) application that you access on demand to expand towards PaaS.

Now, there certainly are a lot of questions about Salesforce's business model, per se, in terms of its very high cost of customer acquisition. The fact is that when you have a subscription-based business, there is going to be customer churn. There's no question about that. On the other hand, they certainly have put this issue on the front burner.

In terms of who is best positioned for all this, I think it's a little too early to tell, because most of the large vendors are only just starting to put their feet in the water. Obviously, IBM, HP, and Microsoft are making moves. SAP has actually had a couple of stumbles on the way there. Oracle has sort of a sitting-on-the-fence strategy.

But, some of the strongest vertical areas would be areas where you are doing planning or testing. Those are clear winners, when it comes to deployment in the cloud, because there is actually relatively little risk factor. When I say vertical, I don't mean vertical industry, but vertical from a software-application standpoint.

Gardner: Who can get kicked in the teeth by this thing?

Baer: Well, Microsoft clearly could get kicked in the teeth, and that's obviously why they've come out with their resource strategy and with their various live-office strategies. Microsoft clearly has the most to lose, because they've been very identified with the rich client.

The Internet Effect

Gardner: Let's go to Brad Shimmin. Brad, we've seen over the past 12 years or so, that the "Internet effect" often seems to move in a winner-take-all fashion, at least for a certain disruptive phase of adoption. Do you think we are going to see the same with cloud computing? Is this a phenomenon that is going to put more power, more impact, and more dominance in the hand of a small number of companies, or is this going to democratize IT?

Shimmin: I'm firmly on the democratic side of this argument, because when you look at the strategy vendors like IBM has, Sun will have, and Cisco has, in terms of how they're rolling out anything that's in the cloud -- whether its PaaS, infrastructure as a service, or SaaS -- they all seem to be doing two things.

One is that they are taking some point solutions that they are going direct with, like IBM with Bluehouse, for example. Secondly, they are going after an independent software vendor (ISV) market. They want to empower folks like, Panorama, Pervasive, Peer1, Mosso, Akamai, Boomi, and all those guys. They're really looking to empower them to go out and deliver services, whether its any of those three layers that I mentioned.

They want to deliver their software and their services, whether it's taking IBM's Lotus software and delivering it directly via some sort of white-label solution on their own, or just mashing up the services that Cisco provides with their WebEx Connect, it doesn't matter. What these companies are doing is allowing this broader feel, allowing this channel of service providers to exist, using their software and their services, and, in some cases, their actual data-center resources.

Gardner: Let me understand. I think you're saying that the organization that can provide the best ecology of partners and provide the best environment to thrive for many other players will do best, whereas, in the past, it seemed that, as an IT vendor, having the most installed base and the most lock-in offered the path to who did best.

Shimmin: Exactly. A good example of that is SAP with their Business By Design software, which is direct to consumer. They launched that a little over year ago and basically pulled back from it. They're getting ready to launch phase two this coming year. That is a testament. Sometimes, you shouldn't do it on your own when it comes to cloud computing.

Gardner: Joe McKendrick, we've seen a couple of impactful bloggers out there saying that the cloud is going to provide this all-or-nothing, and someone is going to come in and dominate the whole thing. Richard Stallman, I think, was perhaps a tip on the arrow for that mentality recently. Now, we've got Brad Shimmin and others saying, "Follow the model. It's one big happy sandbox. Let's get more players in." How do you see it, one or the other?

McKendrick: It's probably going to be quite a diversity of offerings. Cloud computing SaaS is essentially a delivery model and its going to be one of several delivery models. Don't throw out that data center just yet. Most companies aren't ready to commit their mission-critical core applications to cloud computing, especially the larger enterprises. They're staying with the license model at this point.

You see that cloud computing is very much at the periphery of things though. Amazon Web Services is very much a play for small startup companies. WebEx is an example, as Jim mentioned, of a cloud function that's probably used within large enterprises, but, again, it’s not a mission-critical core application.

SaaS cloud computing is one intriguing delivery mechanism, and it will be offered alongside more traditional delivery mechanisms for a long time to come. Probably 20 years from now, we're still going to see data centers on site.

Gardner: This is more of a peripheral activity and you can't dominate from that position. So, I guess you are more on the democratic side.

McKendrick: The democratic side. It's giving a lot of opportunities for new players. In Microsoft, you see something like Zoho coming up with its own compelling office suite. Google also has offerings in that space. I don't think they're going to unseat Microsoft, but it’s a nice alternative and it makes the market more interesting.

Shooting for the Moon

Gardner: Jim Kobielus, I don't know if you're a card player, but there is this game out there called Hearts that we used to play a lot when I was younger. Most of the players would just try to win, but once in a while somebody would look at their hand and say, "It's so good, I'm going to shoot for the moon. I'm going to try to take all the tricks and put everybody out." Do you think there is an opportunity for somebody to try to do that with cloud computing?

Kobielus: I don't think so. I think you hit the nail on the head, Dana, a few minutes ago, when you pointed out that success in the emerging cloud arena depends on having a very broad and deep ecology of partners. I see the partner ecosystem as the new platform for cloud computing, being able to put together a group of partners that provide various differentiated features and services within an overall cloud-computing environment.

Then, the hub partner, as it were, provides some core, enabling infrastructure that binds them all together. Core infrastructures such as, for example, a core analytic environment or distributed data-warehousing environment that manages all of the structured, unstructured, and semi-structured data, manages all of the very compute-intensive analytical workloads, CPUs, and other resources that many or all of the partner solutions can tap into -- a basic utility computing environment.

So if you look at this, then the APPExchange model is very much the template for success in cloud computing. Will others, or can others, provide a similarly rich partner ecosystem environment for their cloud computing efforts? Clearly, well established platform vendors, like SAP, Oracle, IBM, and Microsoft, stand to do very well in the cloud-computing paradigm, because they have substantial, global, differentiated partner ecosystems.

New players who come in to do just collaboration in the cloud, storage in the cloud, and so forth, will not necessarily be the dominating vendors in this new environment. They'll probably be members or participants in several partner ecosystems, providing some core functionality, but they won't dominate to the extent that the established brands will.

Gardner: Brad Shimmin, there is a certain traditional bifurcation in IT vendors. Some focus on platform values, services, and products, monetize around the platform, and are either agnostic or neutral around applications. There are others who are mostly focused on application-level values, monetize around that, and are either picking and choosing platforms or offering wide portability.

Can this bifurcation continue into the cloud era? For those who have an installed base of either platform and infrastructure or applications, can they transfer that to some sort of a cloud-based service offering and leverage off of their installed base?

Shimmin: I think so, Dana, and it's already happening with most of the major players we're talking about -- Oracle aside, because Oracle still seems to be on the fence with this. But, when you look at a company like Microsoft, they seem to be slow to market, and then, once they enter the market, they go really, really fast. They seem to be going really, really fast at the moment with two things, because they have both. They have the infrastructure and they also have the apps. They're going to have both paths.

They have the Azure platform, which is truly a PaaS offering that you use to build your own applications. So it's a layer above the Amazon EC2 infrastructure as a service.

Then they have the full-on SaaS-type products with Microsoft Online Services, which has in it almost the entirety of their collaboration software. So, they have actually sort of leapfrogged IBM Bluehouse a little bit with that.

The point is that these vendors are really looking at their portfolios and seeing which ones fit either of those two models. They're not committing to one or the other, Dana. They're really trying to tackle both ends at once.

Gardner: It seems to me, though, that, if you are focused on the applications, alternative cloud-based offerings for those applications could develop and undercut your business. If you're an infrastructure provider, you could start to offer similar cloud-based infrastructure services. That is to say, you could run your apps, you could perform services, do architectural development of business processes, and cross this barrier between internal and external services, if you stick with a common platform and a de-facto set of standards.

So, do you think, Brad, that there is a disadvantage for being an applications vendor who doesn't move rapidly to the cloud?

Risking the Customer Base

Shimmin: I think you're at a disadvantage. You're at risk of losing your customer base. We're going back to talking about the difference between Microsoft Online Services and Office Live, compared to Zoho and Google Apps, for example. Microsoft can be undercut in a heartbeat, if they don't do this right.

Back to what you were just talking about there, I see it as a distinction between PaaS and infrastructure as a service. There are a number of rudimentary basement layer infrastructure-as-a-service vendors out there who are going to be taking the Amazon model and talking to vendors, as we've been talking about, to act as a part of that ecosystem, and to build out this infrastructure that will allow you to build and deploy any PaaS or SaaS on top of that.

What that says to me is that the vendors who have that strong infrastructure piece -- the infrastructure software and in some cases the hardware and data centers themselves -- are going to be well positioned to play in that ecosystem, regardless of what sort of SaaS applications they have.

Gardner: Salesforce is an example of this. They were very application specific, when they first came out with their SaaS offering, and now they say, "No, no, we don't want to just be application specific. We really want to be cloud platform services specific." Right?

Shimmin: Yeah, they may be following that tail a little bit. If you look at EMC as well, they have sort of a similar approach. They built out a true platform. I think its called the Fortress. It's their own data center. They started with Mozy, which is like a consumer backup, and they have Recovery Point. They just announced a more generic data-center service for that. They'll just keep building out this broader infrastructure as they go.

Gardner: Tony Baer, two points. One, it's almost ironic now that being in the hardware business might actually be a good thing. For years, there was this notion that the hardware is commodity, and there will be no margin. It's the software where all the margin is. If we follow the logic of cloud computing, being a provider of the data center infrastructure, including hardware and storage, could be a safe bet, whereas the margins are going to continually be under assault on the applications. Does that make sense?

Baer: I'd still go along with what Brad was saying. I don't think that just a strict platform alone -- in other words, storage as a service -- is going to ultimately be a predominant play. It's a very niche buy. However, storage, along with what Jim was saying about analytics as a service, atop which you can then build an ecosystem of solution vendors, that's much more a winning combination.

When you look at the cloud, you're looking at a platform play, as opposed to a niche play, with one exception. The exception is that one of the advantages that's been stated about cloud computing is that it gives you a chance to kick the tires before you actually commit to implementing a solution or application. That's the one exception I would make there. Otherwise, I agree with the others here that, in the long run, probably the strongest play is more of a platform-based ecosystem, which will include infrastructure, because you can't have an ecosystem without having underlying infrastructure services.

Gardner: So, we have these cloud providers that are spending, in some cases, billions of dollars on a quarterly basis to build out their data centers. As more and more applications and services move to the cloud, they're not going to be able to get utilization beyond 100 percent. That's all you can get. So, they need to add more servers, storage, and data centers.

As this trend unfolds around the world, they need to be regional. They need to take advantage of lower costs of labor and energy. It seems to me that being a supplier to the cloud data-center providers is a pretty good business right now.

Baer: That is a good business. It's not the end business, but it's a good business. In other words, your successful cloud platform is not going to be just, "I'm going to provide storage as a service or backup as a service." However, to players who are trying to deliver PaaS it certainly is a winning strategy. You won't become a dominant cloud player, but you can certainly play a major part in that ecosystem.

Gardner: Joe McKendrick, let's look at this through the eyes of application development. It seems to me that developers and ISVs, are once again a very important component of how these newer trends unfold. The hearts and minds of developers, where ISVs see their businesses being good, can invest their development in order to get a return.

These green-field application developers, are they going to continue to be concerned about installed-base platforms, or are they going to see more of an opportunity in this PaaS and build their apps of, for, and by the cloud?

The New Heavy Industries

McKendrick: Dana, I just want to go back one point there. It's interesting when we talk about the larger infrastructure providers and cloud providers: the Amazons, and, to some degree, Microsoft. They've become the new heavy industry of the 21st Century. It's no longer the thing to do to pursue smokestack factories, and so forth.

Gardner: These are the new River Rouge build-outs, right?

McKendrick: Exactly. Microsoft, Amazon, and a couple of others are building huge sites, or have built huge sites, along the Columbia river basin, where there are vast amounts of hydroelectric power and cheap energy. We're seeing other communities around the country pursuing these large data center providers. That's become the heavy industry du jour.

Gardner: At a time when they will take any business they can get.

McKendrick: Exactly.

Gardner: So, what about developers, green field versus building to an on-premises platform framework lock in?

McKendrick: Just about every small ISV coming on the market now is offering a SaaS model. This is the way to go with the emerging smaller software-development companies.

For the larger developers, ISVs that are already well-established, it's now another delivery mechanism, another channel to reach their customer base. There are a lot of efficiencies. When you have a cloud model or are working with a cloud model, you don't have to worry about making sure all your customers receive the latest upgrade or deal with problems customers may be having with conflicting software. It's all done once. You do the upgrade once, test it, ensure the quality, deliver it, and it's all done in one location. It makes their job a lot easier.

Gardner: It certainly does. It's a very attractive model, if you're a developer. You don't have to put up a lot of upfront cost. You might not need to go out to a venture capitalist and get $30 million to start your company. You might to be able to do it with an angel or two, or maybe even bootstrap it, right?

McKendrick: Exactly. You don't have to worry about buying and mailing out those CDs.

Gardner: Jim Kobielus, you talk with a lot of startups. I talk with a lot of startups. When it comes to money -- whether it's VC money, angel money, or credit card debt money -- there's a lot of ferment in developing services of, for, and by clouds. I'm not being approached by many startups saying, "We're going to build a business around this Unix platform, this particular flavor of Java, or this particular instance of a Windows runtime environment." Does your view of the market jibe with mine?

Kobielus: Oh yes, very much so. What the whole trend towards SOA started was the gradual dissolution or deconstruction of the underlying platforms, as you mentioned -- OSs, development environments, and the declarative programming languages. This is all buggy-whip territory now in terms of what large and small software vendors are developing to. Pretty much everyone is now developing to a virtualized SOA, cloud environment.

When I say "cloud" in this context, most of the large and small vendors that I talk to -- although they may not use the word cloud -- are really looking at more of a flex-sourcing approach to delivering solutions to market. They might come in with a subscription service, but they often say, "We want to put it out in the market, see if anybody signs up, and then see if from there we want to turn it into some sort of packaged, licensed software offering, or, conceivably, we might turn some aspect of it into a hardware appliance that's optimized for one function."

Most of the vendors that I talk to now have three broad go-to market delivery approaches for flexible delivery of applications or of solutions. They have everything as a service approach, the appliance approach, and the packaged, licensed software approach.

If you look at cloud computing as a Venn diagram, with many smaller bubbles within it, one of the hugest bubbles is this notion of flexible packaging and sourcing of solution functionality.

The "Chinese Wall" between internal hosting and external hosting is dissolving, as more and more organizations say, "You know what. We want to do data warehousing. We'll license a software from vendor X. We might also use their hosted offerings for these particular data marts. We also might go with an appliance from them, for either our data warehousing hub, a particular operational data store, or another deployment wall where the appliance form factor makes most sense."

Gardner: Brad Shimmin, do you see this breakdown too, where the green-field applications, these newer services, can take advantage of ecologies and bring other services into play and create a solution level or a business-process level value? They're going to be all formed by the cloud. They're going to be virtualized. And, they really don't have much concern about what the underlying infrastructure is. Perhaps what they would like to see more of is the ability to run that application in some sort of a coordinated fashion, both on premises and in some cloud.

Pressures on the Market

Shimmin: I do. There are a couple of pressures that are making that so, and I'm very much with Jim on this. When you look at companies like IBM and Microsoft -- Microsoft with their Software plus Services, and IBM with their Foundation Start Appliance, coupled with their Bluehouse software as a service, coupled with their on-premises collaboration software -- you're talking about a solution that spans those three delivery mechanisms.

The pressures I'm talking about that are making that so for the enterprise buyer is that you don't want to have a full SaaS deployment, and you don't want to have a full appliance deployment. When you consider issues like ownership of data, privacy of that data or SOAs, even transaction volumes, there are facets of your enterprise application that are best suited to running in an appliance, in your data center, or in the cloud.

So, these vendors we are talking about here clearly recognize that need, and are trying to re-architect their software so it can run across those three channels in different ways.

For example, let's say you have an email-messaging repository that sits inside your data center. Then, you have the actual transactions for sending out those messages that lives inside Microsoft's Online Services. That's what they're building toward. As I said, they're building toward it. Neither of those vendors -- and they are the two that are leading the way with that approach -- have achieved that. This isn't something you can do today. It's something that you are going to be able to do tomorrow.

Gardner: I think I hear you saying that we shouldn't think of this as an application-by-application decision. That is to say, "I'm going to pick and choose among my 800 applications, and 500 of them will stay on premises, and 300 of them I'll push out into a cloud provider. Let them do it as a SaaS provider. What I hear you saying is not application by application but underlying infrastructure service by infrastructure service, and that's what I will use to divide what I keep on premises and when I go to a cloud.

For example, I might want to keep data services on premises. I might want to keep directory and access management and overall governance on premises, but I might want to pick and choose from a variety of services off the cloud that then create these application processes.

McKendrick: The underlying architecture that a lot of vendors are moving toward to enable that degree of flexible deployment of different form factor -- hosted service, appliance, and packaged license software -- is the notion of shared nothing, massively parallel processing for extreme scale-out capabilities and extreme scale up as well.

In a federated model, where you have different clusters that can be internal, external, or in combinations specialized to particular roles within the application environment, some might be optimized for data warehousing, some might be optimized for business-process management and workflow, and others might be optimized for the upfront delivery, Web 2.0, REST, and all that. But, having shared nothing, massively parallel processing, with a federated middleware fabric in an SOA context, is where everybody is moving their platform and strategy.

Gardner: What's essential for any of that to happen is you need to be doing SOA. You're not going to be able to take advantage of this hybrid model, not just on a application level, but on an infrastructure services level, unless you have already adopted, or well on your way to moving towards SOA. Does anyone disagree with that?

Baer: No disagreeing from me. SOA is the way. Ultimately the vision of SOA -- and this is something I try to bring up -- is that companies are going to be both consumers and providers of services, and the vision of SOA is that it runs both ways. You publish services and you consume services.

We're going to see companies, not necessarily in the IT business, take on more of a role as a provider of service, and they are going to have the infrastructure. They have the infrastructure in place now. SOA connected with grid computing can play a role here, where you have companies providing these services. They're looking for scalability. They're looking to maybe extend their service offerings beyond their corporate walls to their ecosystem of partners, maybe to customers, customers exclusively of that service, and without a prior relationship with the company.

Through SOA, perhaps companies can look at increasing capacity or tapping into capacity as needed in a grid like fashion, either with each other, or with a provider out there such as Amazon or IBM.

SOA an Essential Starting Point

Gardner: I'm definitely telling people that if you like the idea of cloud, if you see yourself moving towards wanting to take advantage of cloud-like services and values over time, you need to get your act together on the SOA front. It's absolutely essential. I think we all agree on that.

We talked about ecologies. We talked about how groupings of companies working together with mutual interdependence, or the fact that they are going to benefit mutually from a certain market orientation, makes sense.

It seems to me that this is not just going to be among the little guys however. There is a certain alignment that might be in the offing among some of these big companies. So, I'll offer up one potential bedfellows opportunity. That would be IBM and Google, perhaps working closely with someone else like Cisco. If you put together what Cisco brings to the table in terms of increasing services as a function of the network and what Google offers, which is more SaaS and PaaS, as well as a tremendous amount of metadata on what consumers are doing, with what IBM does in terms of its touch in the enterprise and in infrastructure, those three together strike me as a powerhouse.

Does anyone disagree, or do we have any other sort of mega-clusters of vendors out there that would align well in a cloud model?

Shimmin: I think, just like you have at the high school prom, some strange pairings and some power pairings. The power pairing you mentioned there is like the prom king and queen, and they will probably be one of those.

There are a number of others that are kind of bizarre, but I think really beneficial, like TIBCO and Cisco, for example. You wouldn't think it, but TIBCO, with their work on Active Matrix as the abstraction and virtualization of software coupled with Cisco's work in the cloud for their own data center that they're rolling out -- and in terms of the equipment they ship into private clouds inside enterprises -- it's a great combination.

Gardner: How about a combination of HP with EDS, and Microsoft plus something else? Does that make a cluster sense?

Shimmin: If you're talking from the SI perspective, absolutely, because Microsoft is channel bound. That's their legacy and their current status. Being able to work with somebody like an EDS would benefit them tenfold in this situation.

Gardner: How about Apple and SAP? How would they line up with a sort of mega-cluster, anybody?

Baer: It would have to be on Steve Jobs' own terms, and I would have a hard time seeing the Germans at SAP go along with that.

Gardner: Oracle, SAP and Apple, is that what you are seeing? One or the other. Apple teaming up with SAP or Oracle?

Baer: When you take a look at Apple's business model, they like to definitely be like the queen bee in a hive of lots of drones.

Gardner: Well, we talked about playing Hearts and shooting for the moon. Isn't Apple a shoot-for-the-moon kind of company?

Baer: Certainly. Take a look at Apple's traditional approach to partnership, and take a look at how they are handling the application space with the iPhone. They will definitely insist on control, and with a powerful player that also insists on control, like SAP, I have a hard time imagining those two coming together.

Gardner: Remember, we're not talking about mergers and acquisitions. We are talking about partnerships.

Baer: I agree with you, but, essentially, when you are talking about partnering in a cloud, it is a form of virtual merger and acquisition.

All For One -- One For All

Gardner: Maybe it is. Interdependency -- we live or die together, all for one, one for all.

How about Amazon? That would be in my thinking a pretty good candidate for prom queen right now. Perhaps there will be some polygamy at the prom, because Amazon could team up potentially with say an Oracle and a Salesforce. Can you imagine such a pairing?

Kobielus: Yeah, because Oracle, a couple of months ago, announced that you can now take your existing Oracle database licenses and you can move them to the Amazon EC cloud and the Amazon storage service. So, to a degree, that partnership foreshadows possibly a larger relationship between those two companies going forward.

I think its really an interesting pairing of Oracle plus Amazon. Once again, I always have to hit the analytics thing on the head, because I think database analytics or cloud-scalable analytics is going to be a key differentiator for most application vendors.

So, out of the blue, I can imagine that Oracle and Amazon partner with one of the leading data-mining vendors, such as SAS Institute. SAS Institute, based in North Carolina, is a privately held firm. They continue to emphasize that they've been doing SaaS as an alternate delivery channel for their very verticalized and content-rich analytic applications for several years now.

I don't think it's inconceivable that Jim Goodnight, the founder of SAS might say, "Yeah, Oracle and Amazon have got a cloud thing going on. It makes great sense for us to take our existing SaaS strategy and bring it into an Oracle/Amazon cloud so we can continue to penetrate a broader and deeper range of verticals with very flexible options.

Gardner: Clearly, in the cloud, the best analytics will have a significant advantage. Now, what about Red Hat? How about a Red Hat-Amazon pairing? Does that make sense?

Shimmin: That's already done. They have both their app server and their operating system running on it now, but interestingly enough, Dana, they're not going to go up the stack any further at this point.

Gardner: They seem to be sort of pulling back. I think you're right.

Shimmin: I talked to them about their enterprise service bus (ESB) in particular a couple of weeks ago, and I was sort of surprised to hear them tell me that they really didn't feel that, given the way their customers purchased software, it was a model that would work for them.

Gardner: I think they've recognized that the virtualized runtime instance with a stripped-down Linux kernel is a really good business and they should stick with their knitting.

Shimmin: Right, but if you look at their portfolio, you have Drools for example and their business process management (BPM) products, for which I can't remember the name. Those two PaaS offerings would be phenomenal. Don't you think?

Gardner: Yeah. Let's go back to Microsoft. Microsoft has an opportunity to shoot for the moon. I'm going to be a little bit of a contrarian on that. They have all the essential pieces. They have a very difficult transformation to make in terms of their business. They have a lot of cash in the bank, and we're in a transformational period.

If you were going to make a big move, now it would be an excellent time to do it. It really comes down to execution -- whether they can get the various feuding parties inside the company to line up well. But, Microsoft also has to make a choice as to whether they want to be everything to everybody or strive for a better ecology.

Does anyone have any thoughts about what Microsoft should do? Should they try to do it all, or should they become more of an infrastructure-focused provider, not try to be buying Yahoo and becoming search and consumer and applications? Leave that to the other players in the ecology Let's look at Microsoft's situation and let's go to Tony Baer.

Baer: For this, I think there's a difference between what they strategically should do as a company, versus what Wall Street would prefer them to do. Wall Street is always looking for quarterly numbers and a show of growth. Obviously, buying something like a Yahoo, even though Yahoo at this point is pretty much damaged goods, provides that obvious growth into an area that Microsoft and Wall Street have been obsessed with.

However, what's smarter in the long run is the whole Software-plus-Services strategy, which is a great idea, but the devil is going to be in the detail. The idea of providing a seamless, or relatively seamless, experience of whether you're working with -- let's say, Word online versus Word at your desk or SharePoint, or whatever -- is a great idea. I think Microsoft is right now puzzling out the technical details, which is that Word online is not going to be the same exact creature as Word on your desktop.

Gardner: Doesn't Microsoft Software-plus-Services put them at odds with the ecology mentality? Doesn't it, in a sense, push these green-field applications that only want to be in tuned with a virtualized environment. Doesn't it turn them off?

Baer: Well, it might, but do you want to follow your customers in terms of how they want to work or do you want to follow a blind ideal. I think what Jim and Brad were saying before is that, in terms of what customers are going to prefer in the long run, it's going to be a mixed bag.

There are going to be certain services that you will want to consume as a service versus some assets, processes, or functions that your corporate policies and matters of governance are going to require that you keep in-house.

Gardner: I think Microsoft has an opportunity to make an offer that developers can't resist -- and probably no one else is in a position to do it -- which is to say, "We will have at least one of the top three clouds. We're going to give you the tools and give you simplicity that Joe the plumber can develop, and we're going to make sure that you have a huge audience of both consumers and businesses that we're going to line up for you." Isn't that a formidable position, Joe McKendrick?

McKendrick: Very much a formidable position. They've already made a lot of moves in this direction: Software plus Services, the Live offerings. They're already positioning a lot of their product line. They work with Amazon and have offerings through the Amazon service as well.

Microsoft gets into everything. Wherever you look, in the enterprise or in computing, they have some kind of offering there. Sometimes, the things don't take off for a while. They sit and bide their time, and eventually it takes hold.

Gardner: Tony Baer that said Apple computer was like a queen bee with drones. We could apply that to Microsoft as well. It might not be an ecology, as much as the queen bee in the hive dictating all the rules and then the drones just click along, making that a pretty good living, but Microsoft makes the lion's share of the dough.

McKendrick: I think that's a good model. In fact, thinking about the Microsoft plus Yahoo, it makes really good sense for them both to be a real powerhouse together in cloud computing. Earlier, I stressed that the providers who dominate the cloud world will be those that focus on extreme scalability, scale out, shared nothing, massively parallel processing being able to sift and analyze petabyte upon petabyte of data from all over especially of the Web 2.0 world especially clickstream information, and so forth.

Microsoft is already very much focused from the highest level on cloud computing with Azure, Live, and so forth. Clearly, they've got all of their online assets from years back. So, they are very much focused on that.

In terms of scalability, Microsoft has one recent announcement that's pivotal to the development of their platform. They acquired a company called DATAllegro a few months ago, a data warehousing appliance vendor whose primary differentiator was a very strong shared nothing, massively parallel-processing architecture that Microsoft is making the core of a near-future SQL server-based data warehousing environment.

The thing is I am fascinated about with DATAllegro's technology is that it can be used to build the underlying scale-out substrate for cloud computing as well -- combining DATAllegro's strengths from a technical side with Yahoo's strengths on parallelization through the MapReduce or Hadoop framework. They've been one of the leaders in pushing to do really massively parallel clickstream analysis on Web 2.0 and social networking information within the Microsoft cloud, I think that would be a killer combination to dominate this world.

Gardner: Why is that important? Microsoft has Yahoo, or at least the search part of Yahoo, and apparently they can just dump the rest, if they want. That gives them all that metadata about what the consumers are doing. Then, they've got all that information about each and every enterprise in small-to-medium sized business that they deal with. They've got the PaaS cloud and their own channel, if not an ecology of channels.

They can go back to those developers and say, "If you want to remain in business, we're the best bet. We can give you the metadata of how to reach the consumers. We can give you the metadata how to reach the businesses. We might even be able to join them together in a transactional relationship. We take a cut. You take a cut." Any thoughts out there?

Shimmin: With regards to Microsoft's channel, as you and Jim were saying, Microsoft is definitely going to be the queen bee and they are definitely going to make it beneficial to this channel to work with them in their cloud initiatives. At the same time, it's also Microsoft's greatest risk.

When you look at their PaaS with Azure, that makes sense for the channel, because how the channel differentiates is by the services they provide their customers directly, and that comes from developing code. But, when you talk about Microsoft's online services, Office Live, and those things, they are in a very precarious predicament of undercutting the values that their channel partners provide.

They're literally saying, "Hey, why do you need a channel partner for the SMB market, just come right to us and give us your credit card, which you can do for a certain number of dollars a month, and you are running."

Gardner: Right, so perhaps Microsoft has the golden opportunity but the transition is perilous, and execution has to be perfect. Just as we had back in the "anti" days, when all of the Unix vendors got together and created what they called the "anti-Microsoft coalition," all these other cloud providers, ISVs, developers, and all the PaaS people are going to get together and try to provide more of a marketplace, in order to if not staunch Microsoft, at least create that democratic approach to cloud. Does that make sense?

Shimmin: Agreed, and interestingly enough -- I can’t believe I'm saying this -- Microsoft has really done something spectacular here, because it all comes back to the developer. What the developer does drives what software you run on the server, in many cases. What Microsoft has done with the Software-plus-Services program initiative, right now, today, using the 3.5 .NET framework in Windows 2008, you can write code that can be dropped in the cloud or on the desktop automatically. You can just write a rule that says, "If I reach a certain service level agreement (SLA), just kick this piece of code to the cloud."

Gardner: So Microsoft and not the business becomes the arbiter.

Shimmin: Exactly

Gardner: OK. I'm afraid we have to wrap this up. We've had an engaging discussion about cloud, but in the context of large vendors, and how the business side of IT will react to this. It's clearly a subject we'll be dealing with for a long time.

I'd like to thank this week's panel. We've been joined by Jim Kobielus, senior analyst at Forrester Research.

Kobielus: Thanks, Dana and thanks everybody. Have a happy Thanksgiving.

Gardner: Tony Baer, senior analyst at Ovum. Thanks, Tony.

Baer: Have a great holiday everybody.

Gardner: Brad Shimmin, principal analyst at Current Analysis.

Shimmin: You're welcome, and thanks for having me. Happy holidays everyone.

Gardner: And last, but not least, Joe McKendrick, independent analyst and prolific blogger on ZDNet and ebizQ.

McKendrick: Thanks Dana, and since everybody will be listening to this in December, have a happy holiday and a happy new year.

Gardner: And I'd also like to thank our charter sponsor for the BriefingsDirect Analyst Insights Edition podcast series, Active Endpoints, maker of the Active VOS visual orchestration system. This is Dana Gardner, principal analyst at Interarbor Solutions. Thanks for listening and come back next time.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Charter Sponsor: Active Endpoints.

Special offer: Download a free, supported 30-day trial of Active Endpoint's ActiveVOS at

Edited transcript of BriefingsDirect Analyst Insights Edition podcast, Vol. 34, on cloud computing and its impact on IT vendors, recorded Nov. 21, 2008. Copyright Interarbor Solutions, LLC, 2005-2008. All rights reserved.

Wednesday, December 10, 2008

Cloud Computing Means More Than Cost Savings: New Models Will Transform Business, Say HP and Capgemini

Transcript of BriefingsDirect podcast on cloud computing adoption, low-risk transitional strategies and innovative business opportunities.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Learn more. Read the related white paper. Sponsor: Hewlett-Packard.

Dana Gardner: Hi, this is Dana Gardner, principal analyst at Interarbor Solutions, and you're listening to BriefingsDirect. Today, we present a sponsored podcast discussion on vision and strategy for cloud computing.

We'll be talking with executives from Capgemini and Hewlett-Packard (HP) on how they see the value and opportunity unfolding around cloud computing, and so-called private-cloud architectures. Many enterprises and service providers are now grappling with how cloud models and economics will impact them.

The specter of a challenging business climate may well hasten the need to seek IT resources that are supported through greater utility approaches to save money, as well as to reach Internet audiences and gain global Web efficiencies.

There are also a host of innovations around the various cloud models that are now just emerging and that we're only beginning to discover. These amount to being able to do business in new ways and using cloud models to accomplish things that simply could not be done before.

The goal is to take advantage of what cloud models offer, but to do so with low risk and in alignment with enterprise IT dictates and requirements around management, security, governance, and visibility.

Here now to provide an in-depth look how cloud models are changing our world is Andy Mulholland, global chief technology officer at Capgemini. Hi, Andy.

Andy Mulholland: Hi, there. Good to be on the call.

Gardner: We're also joined by Tim Hall, director of services-oriented architecture (SOA) products at HP Software and Solutions. Welcome, Tim.

Tim Hall: Thanks, Dana.

Gardner: And also we have Russ Daniels, vice president and CTO of cloud services strategy at HP. Good to have you with us, Russ.

Russ Daniels: Thanks, happy to be here.

Gardner: Russ, let's start with you. We've heard so much about cloud, grid and utility. There are so many different words in respect to these technologies, and these approaches have been around for quite some time, but we seem to be in a state of building excitement.

More opportunity is being associated with cloud computing. Why don't you help us and our audience understand, first and foremost, what is cloud -- and probably just as important -- what it is not?

Daniels: You can argue about what the technology industry is really good at, but I would say the one thing that we definitely excel at is hype. We can take any term and get very excited about it. We start redefining everything that we were already doing with the expectations associated with that new term. We certainly think of what's been happening with cloud in terms of two things that are going on. We think they are both incredibly important to our customers. They both provide great opportunities, but they are quite distinct.

What most hype is really about a year ago we would have described as utility computing. It's this concept of being able to treat all of your resources as pools that you can flexibly bind to the workloads to match up the demands and priorities of the business in a way that improves the agility of the business and also has a huge contribution to the cost structure.

That's important stuff. There is a lot going on there, but it's focused primarily on being able to improve the ability to deliver the workloads that exist in the business today.

We think there is also another thing going. That's what we focus on, when we talk about the cloud, and it's a new model for constructing software. It's a new design pattern, and it allows you to solve problems that really have been out of reach. You can take business needs, which if you tried to address them in the context of traditional IT design and delivery models, would tend to fail or under deliver.

The cloud allows you to go after those problems, to open new markets for the business, to allow it to reach out to customers that it hasn't been able to get to, to improve its differentiation in the market, and to contribute to the real goals of the business itself. That's what we think is exciting about the cloud.

Gardner: Is there something that the people should not confuse the cloud with, something that the people might be thinking it can do that that is really outside the scale of what we are really talking about?

Complexity won't go away

Daniels: There's this vision that's been painted by some in the industry that we're going to flip a switch in a year or so, and all of the existing data centers will just be shut down. There will be these few hyper-scale providers of compute capacity some place remote to the business, and businesses will just plug a wire into the wall and out will flow compute.

That's an incredibly naïve perspective -- that computing will become as fungible and undifferentiated as electrical power. It's certainly true that we can improve the effectiveness, cost efficiency, speed, and agility of IT by taking advantage of virtualization and automation technologies, but we shouldn't imagine that it means that all of the complexities of the world go away.

What's most important for our customers, when they think about this, is not to think of cloud as being an alternative delivery mechanism for everything that IT does today. It's much more an enabler to do things that IT really can't do successfully today.

Gardner: Let's go to Andy Mulholland at Capgemini. What has changed about the world we live in now that makes this cloud vision so appealing and so seductive?

Mulholland: I agree very strongly with Russ that when something new comes out, the industry does tend to get a little over-excited -- but cloud is more significant, because people are thinking again about what they are going to do over the next couple of years with very tough trading conditions.

On one side, there is this premise that it can help me look at how I manage and reduce my cost. Perhaps more importantly, we should say it the other way around. It enables me to address how I deal with a more variable business pattern and pay for what I need when I need it.

Many of the things a business does today are relatively fixed. The enterprise applications, the enterprise resource planning (ERP) systems, we know what they do. The systems they run on are very predictable in their load. Everything works, and that's to Russ's point about how [the cloud] doesn't change everything.

But, what we have is a growing desire and a growing need to find new things in the front office, about how we run our business more effectively, how we get into markets more effectively, and how we trade better. These tend to be small fast-moving projects. They make a very big difference, and we simply don't want the same time scale in provisioning for them.

Increasingly, probably over the next couple of years, people don't want to spend capital on them. They'll want to pay for them operationally. They represent a new market, a new technique, a new set of standards, and a new set of technologies. All of that comes together in where cloud is going to go and make the difference to businesses.

So, it's an interesting time, when what I want from technology in the business is shifting to more porous, across-the-firewall business on the Web, and at the same time I desire to manage how much I spend, how quickly I spend it, or whether I can attribute cost directly to how much I need when I need it. It's being driven by more care about the financial side.

Gardner: Let's go to Tim Hall. When we talk about this ability to have elastic resources, provisioning IT resources as they are needed, and reducing costs and waste as a result, we are really getting more toward an economic story. What are the root problems that we are trying to solve here? What is it about the way IT is done now that needs cloud?

Hall: There are a bunch of things, Dana. I can't tell you how many customers I've visited in the last 12 to 18 months who have told me they can't install another server in their data center, and the reason is that they are out of power and cooling.

There's been a lot of talk recently about energy efficiency and this trend toward painting the world green. A lot of the new hardware appliances and other things that are coming on the market are consuming lower power, offering better cooling technologies, and have advanced automation capabilities. We can do things with managing those systems.

There's where virtualization technology come to play. We can use these pieces of hardware more efficiently, and all of this is trying to contribute to lowering the complexity of what we've currently got. The two things that you continue to see IT wanting to do are reduce cost and reduce complexity. The ways in which they're doing that come at it from two different dimensions. One is an operational dimension, and the other is from this generation of applications. A technology refresh is happening in both of those areas.

Interest in 'private' cloud

Gardner: Andy, we have heard a quite a bit about private cloud, taking these architectural advantages and not necessarily going through an outsourced provider, but building them out on your own in your organization. What are you seeing in terms of the interest in the private cloud for enterprises among typical customers of Capgemini and HP? How does that differ from what they might want to do with the more public clouds? What's the breakout? How much interest in either public or private?

Mulholland: If you're going to do it internally, it's much more tied up with the way you're running your business and probably around the way you're adopting social software and collaboration tools. I often talk about it in the terms of a distributed business. If we look at what's happening to business today, the old adage was that we will sell more of less, we will reduce the numbers of lines, try to sell more of each line, and that will make us able to optimize the process, and do all the good things we know.

What's happening today in most markets is that businesses are seeing quite the reverse. They are selling less of more. They have to have more products, more variation and better tailoring for niches to win new business, new share, and better revenue -- at better margins.

When you move to that model, one of the challenges is how to support it. If I'm only going to have a few product lines, and I am going to run them for three years, I can train people, build knowledge, and operate in a different way. Instead, what we are seeing is, as I move to the other model, I need a lot more flexibility in the way I find the person who knows how we share and build information, etc.

The whole idea is social software. That means provisioning in a very different way, and that leads people to consider how to do that, particularly in distributed businesses. Is there a more effective way than having hundreds of different instances in different places, hundreds of appliances, or whatever you might think of?

So, in one direction, we see that trend. In the other direction, we see a trend where people want to sell or consume services from other people or sell to other people as businesses. It's a bit like with mashups. Everyone always points to the housing map, but the other one to point to is the number of people who use some version of Google Maps on which to build something. That's making it available to be used. Cloud services can be delivered in that manner, so people on the Internet can consume and buy services from you to blend in with what they want to do as well.

When you look at that argument and you look at what's happening, you start to recognize there are already a number of very well known brands that sell through the Internet and combine their services, which are effectively doing this already. The challenge in this is how it moves from being something that a handful of Web-based businesses are using. How do more businesses learn how to exploit that market and take their share of commercial revenue from that market?

Daniels: Can I just expand a little bit on what Andy said here, because it's probably the most important thing for people to work through? There are really these two things going on, and there is a relationship between them.

One is simply how do I deliver my existing workloads at lower cost by taking advantage of virtualization and automation? I might do that in my own data center, and many of our customers are in fact doing that. As I do that, it also gives me the flexibility to get those same kinds of resources externally, and that could be an advantage through some variable cost perspective.

But, there is this other thing going on that's really critical. It's not simply a matter of how infrastructures are architected. Whether you have an internal utility or external utility, it's much more how you design software and what kinds of problems you are solving.

When we think about the cloud, we don't think it's just a matter of how infrastructure is packaged, but it's really a combination of the impact of service oriented architecture (SOA) starting to break apart applications. We think more about the services to separate out the data from the applications, so that you can get at the data without having to go through complex application integrations.

That's one major piece. There's another piece around taking advantage of Web 2.0 innovations, which includes both how you can create rich user experiences in the context of browsers in these remote execution models, but also significantly it's the social dimension. How can you take advantage of the innovation that's occurred in the consumer space by understanding the importance of bringing people together?

Scaling up, scaling down

Finally, all of that is being enabled by a design pattern that allows these kinds of workloads to scale up, scale down, and be able to handle huge amounts of potential demand, but do it in a very flexible and economic way.

It's the combination of all those things together that allow businesses now to start to use technology to solve problems, to create advantages, and as Andy was saying -- particularly in these uncertain times -- to tackle these problems in a way that lets them move where the opportunities are. It lets them experiment readily and to try things, without finding themselves in huge, long-term commitments when some of those experiments fail.

Gardner: It sounds, Tim, as if cloud computing is getting an advantage from a psychological point of view. It's changing the way people think about IT and application development. It helps us understand how this thinking needs to move into the enterprise. Who at the enterprise level is in charge of making sure that cloud computing happens properly?

Hall: That's exactly where governance comes in. This can't be a free for all. The question is how you organize, govern, and provide guidance for what's available. What has been looked at? How do we handle issues of compliance, especially as you get into some more interesting regulatory requirements?

We're finding that certain data cannot exist outside the borders of a particular country, and so it can't be a free for all. You have to establish a culture of governance and build up processes and procedures within IT. How are you going to tackle the policy complaints issues and some of the consistency checking that are going to need to go on?

One thing Andy mentioned at the top, which is very interesting, is the charge-back model and the variable costs that go along with this. Are you sure you know what those costs are? Do you know where they break in terms of scale? Do you have control over who is gaining access to these resources, so it doesn't become a free for all?

If you have two organizations entering into a trading partner agreement for some kind of cloud-based services, and you're distributing the use of that service more broadly within your organization, who is responsible if somebody is violating the terms of use? As Russ said, it's establishing these policies and procedures and formalizing them in a way that IT can effectively be in control of them to take advantage of these opportunities. I think it's critical.

Mulholland: An interesting point you made there Tim is that you really stress the challenge that says we have a very variable business model. Everybody is getting more and more into the variable business model, and it's very difficult to stay in control. It's very difficult to attribute cost to the various diverse activities, and it's very difficult sometimes to look the auditor in the eye and say you actually knew how all of this worked.

These are new challenges. Let's get back to Russ's point. We've got new challenges here, and what's happening is that it's almost happening ad-hoc. In many companies, they're trying to exploit these things, but they are doing it with a complete lack of structure. By bringing in a cloud model successfully, you're actually introducing some structure to support the very activities that people are increasingly experimenting with in their businesses today.

Gardner: This notion of successfully implementing cloud models certainly seems top of mind for organizations. Russ Daniels, what's the first step? Is this an organizational shift, a mindset shift, technology, all the above? How do you get a handle on making this a successful transition?

Daniels: If you think in these two dimensions, every IT organization struggles with simply delivering on the commitments that they have today, and every organization struggles to free up money to innovate and deliver new business value. When you think about the opportunities to take advantage of infrastructure as a service, it's important to sort through those services that IT is delivering to the business and understand which ones are best suited and are most likely to be able to be moved into those forms to drive down cost.

What we find with our customers is that many workloads are important to the business, but they are not mission critical. In many of these workloads, good-enough delivery is good enough.

There are other workloads that are mission critical. They are the things that when they go down, the business goes down. Those things you have to put a huge amount of focus on and deliver at the highest possible quality. So, distinguishing between those types of workloads, identifying those where good-enough delivery is appropriate, and moving those into virtualized and automated delivery models, positions you to take advantage of external infrastructure capabilities as appropriate. That's one side.

The key challenge

The other side is one I think all of us have always realized. The key challenge for any IT organization is to understand what the business really needs, where the business value is, and how technology can help deliver that. This question of business-IT alignment is always the heart of the problem, and it will be certainly be true in terms of how the business chooses to go after cloud-based opportunities.

We think the cloud is great for connecting. It's great for connecting business to business. It's great for connecting business to its customers. It's great for connecting people to people. It's great for connecting the experiences that people have, as they move through their day and the changing circumstances they find themselves in.

All of that connecting is enabled in the cloud, because the cloud provides a persistence. It's a great place to capture state, because the services exist over long durations, and they have pervasive access, so you can get at that state and the context related to that state. It's those capabilities that make the cloud so great for connecting.

Where is connecting important to your business? That's ultimately a business question, not a technology question. The focus should be on having people who can map from what the business needs to understanding how to exploit this new expressiveness that the cloud brings to solve the most pressing challenges, or to exploit the most exciting opportunities that the business faces.

Gardner: Andy, as organizations on the business side recognize how to take advantage of these connections of doing business differently, vis-à-vis the cloud and other partners, they are going to come back to the IT department and say, "Now enable it."

Is there an advantage for those organizations that already have embraced and embarked on a SOA journey and who have implemented governance and managing services internally? Are they are going to be in a better position when those business people come and ask for them to get cloud oriented?

Mulholland: It's pretty logical, when you follow through what we've been talking about. If you're talking about connecting to and using an environment largely based on services, whether you put a big “S” or a small “s,” i.e., business services, or technology services in SOA, it's pretty obvious that if you have no way internally to relate to that, you're going to have a problem.

The good news about this conversation is that we've talked a lot about the new world and new challenges, and the things people are starting to do, which involves that new world. We carefully separated the idea that the old world is immediately going to jump into this.

The point about that is, if you have been doing new stuff, and you are building new stuff inside the organization, you really ought to have started doing that around SOA. If you're using services correctly internally, then of course, you can cross the firewall and start to use services outside, and blend them together.

The big question is how people think about deploying SOA internally. Some months ago, Russ and I were discussing this. We felt that there was a serious disconnect in people's understanding. For some, it's "Oh, let's try this project with SOA. I'm not actually recognizing it as more than a project, or recognizing it as a significant move inside the business, addressing this new generation of fast moving, fast changing things which are much more in the front office, much more likely to involve the Internet and the Web in some way or other."

Most of the SOA people were thinking about what I called "EAI 2.0." It's just a better way of doing technology integration. Some companies have grasped the idea that it's about doing better business and putting the business costs together in a different way. But, it's still quite a tough issue to address. I'm sure Tim would have some pretty good war stories about how that issue has played out with the things he's seen, as well as some things I have seen in that space.

Gardner: How about that, Tim? How well are organizations that have not necessarily embarked meaningfully on SOA positioned to take advantage of cloud?

Hall: The really interesting point is who is driving the initiative. So some of the things that Andy mentioned are being driven from the bottom up. Folks are looking at this as an integration technology, instead of a complete transformation of how they deliver service orientation or business services more comprehensively and more flexibly to address some of the unique challenges that the business is facing. And of course, they're asking IT to do more with less and better faster.

Four important things

You're not going to do it using the same old technologies that you had in your bag. There are four important things that we're learning about how to do things better as we move forward in the IT landscape. SOA adoption, as a transformational agenda, is a microcosm of some concepts that apply very specifically to cloud and preparing people for cloud adoption.

The four things are, first, once you start to move into these loosely coupled technologies, you have the opportunity to do intermediation, and that intermediation can largely be transparent. What that means is that you have an opportunity to do things like compliance checking for such things as information that shouldn't be leaving the firewall, for example.

The second thing is that you have the opportunity to invest in and capture significant amount of metadata about the things that you are using, be it things that you built or things that you are consuming from a third party. That leads to the ability for you to do more in terms of automation.

The third thing is the notion of formalizing the relationships between the consumers and providers of these capabilities. I think of this as a volume control, if you will, where you do want to capture, at a minimum, the fact that somebody is using some service, be it inside or outside the four walls of your data center. Depending on the relationship you have with that person, you may want more structure. You may want more formalization and structure in that agreement, all the way up to what Andy and Russ mentioned around charge back.

Finally, it's just the whole notion of moving away from capability centric IT and moving more towards service orientation. We shouldn't be worried about fan speed and CPU power. We should be looking at whether we have a capability and how flexible is it for us to deliver that at any sort of scale.

This goes to Russ's comments about how we identify effectively, from the top down, what's really core to the business and what's going to drive and fuel the business activities, versus what is contextual and can be delivered at some level of quality, but simply isn't core to the central focus of the organization.

Gardner: Being involved with social media, I encounter lots of comments. There is plenty of dialog going around, but it seems clear to me that there are still plenty of naysayers out there about cloud, particularly for companies that are considering putting certain applications and services or data up in the cloud.

Time and time again, I hear people saying they will never do that. It's not secure. Russ Daniels, what are some of the boundaries here as to what should or shouldn't be brought into the cloud, at least in a medium or short term, in terms of IT functions and application set?

Daniels There is a set of low-level concerns, when you think about a workload that an enterprise depends upon. There's a set of requirements associated with that workload that includes things like security, which is an obvious one. There's a need for compliance, so can you satisfy your auditors that, in fact, you are managing your proprietary information appropriately. Those kinds of things you can understand as being requirements for any delivery method that you would consider for that workload.

We have to saw through that. Certainly, many of the infrastructures of the service offerings that exist on the market today are relatively simple and can't satisfy many of those demands that enterprises have. That's one of the reasons why you see the uptake for these things frequently happening at the level of the department in an enterprise, where either they don't have the sensitivities, or they lack the awareness of those business risks.

That's one piece of it. There is a different angle, though. The cloud allows us to go after problems that we haven't been successful in addressing before. To be successful doing that, we have to design things. We have to get the design right, as Andy was saying, because service orientation is at the heart of this. It's around how I understand the services that facilitate the goals and objectives of my business. Understanding those services in the context of the business is absolutely critical.

Too much of what's happened in the SOA space has been limited to technical enablement. That's absolutely critical, but isn't sufficient. You have to also understand from the business concerns how to capture the key roles and responsibilities of the business. How do I understand the information that's necessary for those roles to fulfill those responsibilities and where that information needs to be exchanged? That's what I can model as the service. Then, I'm modeling those services in the context of business concerns about implementation.

Getting that picture in your head, and then taking advantage of the cloud, allows you to solve these problems in a way that isn't limited simply to what you can do within your own business. It can be extended across your supply chain. It can be extended across your channels and customers, so you can address more broadly the ecosystem in which you operate. The cloud lets you do that.

Interactions and transactions

Mulholland: Can I pick up on this note? As you can probably tell, we've all worked together on this and share it. One of the keys in this, when we have talked about it with people -- to pick up some of Russ's point -- is the difference between interactions which is a lot of this new market, and transactions which was the old IT market.

When you look at any IT system, it's fundamentally about getting a safe transaction to record what you have done. But, if you think about someone trying to decide what they're going to buy from you, like buying an airline ticket, deciding which flight and how much money they're going to pay and which extras they're going to have, it's a lot of interactions. The speed at which interactions go back and forth isn't that critical if it takes a half second longer.

To get back to Russ's point that he made very clearly, it's something different. State is a big issue in it. Data is not the same issue in the same way, as we are used to seeing with applications. When you start to see it that way, you start to understand why we are using a cloud-based service in this way. It's a perfectly acceptable commercial risk-reward, or whatever word you want to think of in terms of, "Is the service-level agreement (SLA) good enough to choreograph this?" We are trying to just put something different here.

Gardner: Andy, here's a follow up. The cloud naysayers, have they got some things wrong?

Mulholland: I'm not sure they've got things wrong. There's a huge temptation -- I have to avoid it personally, and I am sure everyone else does -- that when you're shown something new, you try and apply it to what you have already, and you try and bend it to fit. Much of what I hear from the naysayers is the assumption that the cloud is about applying it to the current generation of IT, and some of the issues that we touched on earlier.

Actually, it's also about understanding that you have a new set of challenges, a new set of requirements from the business, and therefore, these are not easily addressed in the old way. You need to change what you are doing. It's not so much that they’re wrong, but that they're looking at the wrong thing as the target for where they should be applying the idea of how to use cloud computing.

Daniels Andy hit it exactly right. Frequently, when people think about what they are trying to do, they think in the context of those existing services that they deliver to the business, and many of those are, in fact, transactional. We don't think the cloud is great for “transactionality,” for deep, technical reasons.

In the cloud, you need to be able to scale arbitrarily and you need to be able to do that where you get linear increase in throughput, as you scale out horizontally, which says there is a huge dependency on being able to work concurrently and to work in parallel. Transactional workloads don't lend themselves to that.

It's very difficult to fan out the transactional workloads, because ultimately that item can only be taken out of inventory once, and so you have to bring everything back together. The two-phase commit design pattern isn't very well suited to horizontal scaling. So, the place where the cloud is great is where you're not focused on supporting transactions, but interactions, where you are connecting. It's being able to take state from participants in an extended supply chain and propagate that information up through data feeds, up into a cloud service.

For example, that information might be related to the carbon footprint related to material flowing through an extended supply chain. Each of the participants in an extended supply chain can simply publish a data stream that captures the carbon footprint of the materials that they will be producing. Now, you can run analytics in the cloud, using search-like algorithms, to answer questions about the carbon footprint for some end products. You don't have to do the detailed process integrations. You don't have to provide detailed transactional integrations across the supply chain system to support it.

It's exactly that new expressiveness that allows us to go after problems that we really couldn't have done affordably in the past. Because we couldn't do them that way, we ended up doing things manually and in emergencies. If you think about product traceability, it's the same problem, very difficult to deal with from a technology integration perspective in the traditional ways. As a result, when there's a problem, we have people pawing through information spreadsheets manually and providing the answers too late to be helpful.

Gardner: Given that we're dramatically changing the way we're doing things in order to take advantage of these efficiencies and new capabilities, do we need to create a new hierarchy or metric organizationally?

That is to say, the role of architect now needs to include someone who sees this all through a cloud perspective. Do we need a higher order architect, perhaps at a services level? I guess the question is, how does IT and/or that intersection between business and IT change in order to take advantage of this properly?

Mulholland: I'll have a shot at this. It's something that we've been tracking with some interest, because we focus much more strongly on the business and to how the technology is used. We rely on HP to give us the products to be able to marry up to that business requirement.

If you track backward through this, in 2006, Tim O'Reilly tidied up his views on what Web 2.0 is and how it works. He started to add to it ideas of a business model of how business using it might be different.

About three months after that, Andrew McAfee at Harvard coined the term Enterprise 2.0 in Harvard Business Review Online, and started to build the idea of a business-model innovation. In other words, your business model could be different to present products, deal in markets, do supply chains, and the kind of things Russ was just mentioning.

About three months after that, Forrester Research produced a viewpoint, which said that this should be treated as a different branch of technology, and they called it business technology. Their argument was that, at the time when the mini computers existed around the late 1980s, and PCs and networks were just appearing, the term "information technology (IT)" was first coined to separate off a new cluster of technology that was used in a different way, i.e. around personal computing, centered on information, not on big mini systems, and transactions, and so on.

The term IT was used to describe a change, which if you think it through and remember those times -- and some of us still do -- brought us to the viewpoint that we had a different set of technologies. Applying those technologies led people to coin the term business process re-engineering, let's rethink how it works. We developed a whole new model of client-server, and finally and not least a whole new generation of ways of managing and controlling the business through enterprise resource planning.

The argument that we're extending is this: the topic we're talking about -- the cluster technologies, the role they play in a business, how you build, and deliver, and maintain them -- is a different branch with different skills and activities and therefore it should be called “business technology.” Whether you believe the argument or not is a different question, but it's very interesting that one of the foremost industry analysts actually came up with that proposal about 18 or 20 months ago.

Gardner: It sounds as if we have a transformation that needs to take place at multiple levels within these organizations. Let's focus for a second on for those companies that get it right. They can make the transformation, take advantage of these newer models, extend their boundaries, and be more into interactions, and what that entails. What's in it for them? Let's go first to Tim. For organizations that get this right, what's the payoff?

Hall: Go back to the three basic things that we've been talking about, which are decreased complexity, increased agility, and lower cost. They're the fundamentals of a business. As technologists sometimes we get too enamored with the buzzwords and the hype that Russ Daniels was mentioning at the top. We forget why we are doing this in the first place, and it's very simple. It's to take advantage of and use this business technology as a strategic weapon, while at the same time lowering cost, lowering complexity, and increasing your new business agility.

Gardner: Russ Daniels, same question. For those that get it right, what are the payoffs?

Daniels Tim's list is exactly where the focus is, but I would say again that the cloud allows you to deliver the business results that matter. In other words, it really has to be thought of in the context of IT’s technology for business, and the key business challenges that we see our customers facing today are how to develop new markets? How do they take advantage of the abilities they have and deliver them to new customers? How can they understand better what their customers need, and how can they fit in and connect with them?

The cloud provides great capabilities for that. We think that it's still early, and you can see the promise in things like the recommendation engines that you find at online shopping sites. You're searching for something, and, based on your buying history, your demographics, your search behaviors, and then comparing that to the behaviors of others, the site can provide you with suggestions about other things that might be of interest to you as well. The technology helps identify your intentions and then offers suggestions to help you find things better suited for your needs than what you could have expressed or identified yourself.

That's a wonderful opportunity, and to be able to expand that approach into more and more of the ways that a business connects has huge implications. So, it's an upside opportunity. It's enabled by the agility and by the lower cost, but the key thing is that it allows you to open the markets, to differentiate your offerings, and it allows you to improve profits or gain market share. Those are the things that businesses get.

Gardner: Andy, the last word to you.

Mulholland: Relatively speaking, [cloud computing] is unstoppable. The question is whether you'll crash into it or migrate into it. Why is it unstoppable? Because we're watching a business shift, people have to find ways to compete better in the market. Much of that is around. "How do I add smart services? How do I make products more available? How do I communicate directly and intimately with people, so they know what they want to buy from me?" All of those things are already developing in many businesses today, and people are building solutions to do that, sometimes gracefully, and sometimes not at all gracefully.

In other words, just as we had with the PC, where we basically were driven into it, some companies got there in a very ungraceful way and had to figure out afterward how to sort out the mess. Others did have a strategy, and emerged in a very graceful way. I think we're in the same situation. Users wanting social software have taken us there to run and do things better. We've been taken there by businesses needing to get into new markets.

The challenge for the CIO and the IT department is, "How will I enable that to be a graceful migration," not "How will I wait until it becomes a real issue, and then do something about it?"

Daniels: The opportunity that exists is for those people who break out of their traditional mindset around transactions and really start thinking about what this opportunity is in front of them, and how to use this innovation as a weapon. Those are the ones who are going to see the biggest benefit, because they will be able to take advantage of it more quickly. We've seen this with lots of the technology waves that have come before. Those early movers, who can break out of that traditional mindset -- whatever it was at the time -- to the next technology disruption, are the ones who see the biggest benefit.

Gardner: I'm afraid we have to leave it there. We've been discussing cloud models, impact, direction, and strategy. Andy Mulholland, the global chief technology officer at Capgemini, has joined us. Thanks so much, Andy.

Mulholland: Thank you, too, for inviting me.

Gardner: We were also were joined by Tim Hall, director of SOA products at HP Software and Solutions. Thank you, Tim.

Hall: Thank you, Dana.

Gardner: And also, Russ Daniels, vice president and CTO of cloud services strategy at HP. Thank you, Russ.

Daniels Thanks again.

Gardner: This is Dana Gardner, principal analyst at Interarbor Solutions. You've been listening to a sponsored BriefingsDirect podcast. Thanks for listening, and come back next time.

Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Learn more. Read the related white paper. Sponsor: Hewlett-Packard.

Transcript of BriefingsDirect podcast on clouds computing adoption, low-risk transitional strategies and innovative business opportunities. Copyright Interarbor Solutions, LLC, 2005-2008. All rights reserved.