Sunday, May 03, 2009

BriefingsDirect Analysts Unpack Platform as a Service and Measure Future Impact on Enterprises

Edited transcript of BriefingsDirect Analyst Insights Edition podcast, Vol. 40 on the promises of platform as a service and its apparent lack of traction among enterprise managers.

Listen to the podcast. Download the podcast. Find it on iTunes and Charter Sponsor: Active Endpoints. Sponsor: TIBCO Software.

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, Volume 40. I'm your host and moderator, Dana Gardner, principal analyst at Interarbor Solutions.

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. We also come to you through the support of TIBCO Software.

Our topic this week on BriefingsDirect Analyst Insights Edition, and it is the week of April 13, 2009, centers on platform as a service (PaaS). This is the part of the cloud taxonomy that pertains to software and services development over the Internet.

Developers can use tools and testing apparatus as a service on someone else's data center, building and refining their applications and services, and even putting them through some real world performance testing, before going into production and use.

Where these services finally find a home or a runtime platform of some kind can come in a number of ways. Perhaps they run on the same cloud they were developed on. Perhaps they move to other clouds, on-premises servers or data center, or even other kinds of hosting models -- or maybe all of the above options in some form.

We've seen a great deal of interest by the developers in using Amazon's Web services for PaaS. We've also seen a lot of action and interest from the likes of Google, Microsoft, and even IBM. Of course, there are quite a few ecology players that have been out there for sometime as well.

PaaS is not passé. Yet a lot of enterprise operators and mangers that I speak to don't even see PaaS on their radar. They're much more interested in the infrastructure-as-a-service (IaaS) components of cloud computing and software as a service (SaaS) to a certain degree, but they're not quite sure about PaaS.

So, what is PaaS good for and how will it impact enterprises? That's the question we're here to help you understand. What is the reality and where will we end up in regard to PaaS? We're joined by a panel of guests and analysts to help dig into the enterprise role of PaaS.

Please join me in welcoming our analyst guests this week, Jim Kobielus, senior analyst at Forrester Research. Hey, Jim.

Jim Kobielus: Hi, everybody.

Gardner: David A. Kelly, president of Upside Research.

David A. Kelly: Hey, Dana. Glad to be here.

Gardner: And Mike Meehan, senior analyst at Current Analysis.

Mike Meehan: How you're doing, Dana?

Gardner: I'm doing well. I also have several guests I'd like to introduce. Joining us today is Jonathan Bryce, co-founder of Mosso at Rackspace. Welcome to the show, Jonathan.

Jonathan Bryce: Thanks for having me, Dana.

Gardner: And Rourke McNamara, product marketing director at TIBCO Software. Welcome.

Rourke McNamara: Good morning, Dana.

Gardner: Let's start with Jonathan Bryce at Mosso. You saw what was possible in terms of development of websites and web applications quite sometime ago. Tell us something about Mosso, what you do, how you developed the notion of PaaS, and where you think it's heading.

Bryce: I started Mosso with a co-worker, Todd Morey, and we started developing it in 2005. We were working at Rackspace hosting, which is a large traditional managed hosting company, and they agreed to fund this idea and help us get it off the ground. We started Mosso out of a need that we had. I'm a developer. Todd is a designer and a marketer. We realized that what we really enjoyed working on, and where we really felt like we could add value, was in building applications.

If you built an application of any complexity or significance at all, then you ended up having to take on the hosting burden as well. That required server and network administration, and making sure that you were on top of the security, the patching, the backup, and all those things that come along with it.

We really wanted to find a way to create a technology offering that took care of all of the headaches around hosting an application, so that developers could focus just on writing the code. That was the premise that we started Mosso with.

Gardner: That seems to have attracted developers, but is this something that you see a wider embrace of, or understanding of, the inside of enterprises?

Bryce: When we started out, we definitely were targeting people who were similar to us, who were independent or small development and design shops. Over time, we found that this is a more general-purpose tool that did appeal to a broader audience.

We also realized that when you're working at the PaaS layer, you have to productize. So, you have to standardize on technology and on what's

We also realized that when you're working at the PaaS layer, you have to productize.

possible to do in the environment. Within the last year, we've expanded our product portfolio to include a couple of IaaS offerings as well.

One is called Cloud Files, and it's cloud storage. Another is Cloud Servers, and it's virtual computing. Within the enterprise -- it may be a marketing site for major a sporting event that's doing some special promotion -- we're seeing people who don't want to have to run all that infrastructure. They don't know how popular it's going to be, so they'll use it for a small piece.

We're not seeing a lot of enterprises moving wholesale their SAP systems and all of these big enterprise software packages to the platform. We do see them using PaaS to augment their existing technology for new projects that they're working on.

Signs of movement?

Gardner: These new projects, are they typically a certain type of development environment -- a scripting environment, for example -- or are you seeing more movement out towards some of the more major development frameworks?

Bryce: It varies. It depends on which company it is. Some of them want to build it themselves from scratch, and they will write it in .NET and start without any type of a framework. But, we've also seen a lot of pickup in the open-source content management frameworks like Drupal. Wordpress is now used for a lot more than just blogging. We've seen a lot of times where people have started with a framework like that and then gone from there.

Gardner: Okay. Let's go to Jim Kobielus. Jim, we've seen this many times in the past, where developers grasp something quickly and early and end up driving it into much more mainstream adoption. Do you expect that that's going to also be the case with PaaS?

Kobielus: I think so. It's the early adopters who are looking for
a platform where, first of all, they don't have the internal resources to go out and acquire the hardware and all the infrastructure, and license the application servers, the tools and everything they need to get up and running out a new project.

Just as important, from the very start they want 24x7 reliability from whatever platform they plan to execute this code on. So, they're looking for a strategic partner who can handle all the hosting and, to a great degree, can also provide them with the tooling they need, both for the development and ongoing management of the code.

The early adopters have big plans, but small resources. The public-cloud hosting of these applications is important for those applications, where, as a previous speaker indicated, there might be a strong orientation toward business-to-business (B2B) applications, or delivering it out to the broader Internet as a service. They want something that's already out there on the cloud that will be immediately available and accessible worldwide, and with the reliability to back it up.

Gardner: Do you see an opportunity for these tools and services to play a role in how data warehouses are being utilized, perhaps some mining or some additional tooling that you might use across different domains -- a data set within your organization, perhaps something you've leased or rented for a period of time, maybe some publicly available metadata. Wouldn't PaaS make sense, when you're doing these data chores that involve multiple data sources?

Kobielus: Very much so. In fact, a lot of data mining projects are like that. First and foremost, data mining projects depend on discovering sources of data, both internal and external -- customer data, market intelligence data -- and then being able to extract, transform, load, cleanse, and consolidate the data into an analytical data mart.

I am using that term in a logical sense. The analytical data mart then can be mined to do rational analysis and scoring. So, the analytical data mart

All that clickstream mining is happening. That's a core of the Web 2.0 business model

for data mining projects can become fairly large fairly quickly, with all of the different sources of data being brought into it.

With the whole Web 2.0 phenomenon, more and more organizations and businesses are making their entire business model dependent on their Web presence and on their ability as a business to continue to mine the clickstream data user sessions to be able to identify patterns related to the customer experience and preferences, so that they can then up-sell customers, or hold onto the customers.

All that clickstream mining is happening. That's a core of the Web 2.0 business model, and that depends on a big, old clickstream data mart being maintained. As I indicated earlier, small businesses that are going down this road are looking increasingly for a clickstream data mart in the cloud that can be managed -- and it might be hundreds of terabytes or even petabytes -- that can be provided to them fairly inexpensively, and fairly quickly through the cloud model.

That's an example where PaaS is the underlying platform for the analytical data mart in the cloud for clickstream analysis for the whole Web 2.0 business model.

Gardner: Interesting. Let's go to David A. Kelly. David, you look at business process management (BPM) quite a bit. Are there any aspects of business process, putting them together processes that are going to hop across multiple services with multiple sources? Does PaaS relate to that?

Role for process automation

Kelly: Absolutely. There's a big role for business process, process management, and process automation that will be able to cross boundaries here and access different processes across cloud. So, there is an opportunity for PaaS to incorporate those types of capabilities. We're just seeing the beginning of that in terms of what's out there, and over the next couple of years, we're going to see more focus on that, as organizations start to deploy more applications, and develop more cloud-focused and cloud-based applications.

The need will be to manage and automate those processes across different boundaries. Just as Jim said on the data warehouse, business-intelligence side, we have the same thing on the process side. We're going to see development of more tools in that area and capabilities and services that allow organizations to do that.

Gardner: What's interesting about what you're saying and what Jim said is that these aren't necessarily cost-saving measures. We think about cloud as being a way to reduce cost and waste, increase utilization, take advantage of virtualized fabric of infrastructure services, and so on.

You've got to talk about being able to do something quite different that hadn't been able to be done before, because you have elevated yourself to the cloud. Doesn't this whole development in the cloud smack of that as well -- not just cost savings, but being able to do things that hadn't been possible before?

Kobielus: It's rapid, cheap provisioning in the cloud. It's the bubble phenomena, and quickly the bubble grows large like a data mart, because it can grow quickly in the cloud, and then it quickly gets de-provisioned.

Gardner: Dave?

Kelly: I agree with that. Dana, what you bring up is this idea of more nimble processes, and more nimble response -- faster, more agile response to changes. In that regard, from the process standpoint, it's like you're looking at more nimble process management or nimble BPM solutions that we're going to see down the road in order to be able to respond more quickly to business changes.

Gardner: What about that Mike Meehan -- this idea of nimble activities as a result of moving some or significant portion of what you are doing to the cloud? Does that make sense to you?

Meehan: It does, but it depends on what you are talking about as what you want to enable the nimbleness for. We're really talking about business processes that involve third parties and outside parties here. When you're still behind the firewall, it's not going to have as much of immediate impact. What we're really talking about is that we've got a cloud-based runtime out there.

Now, we're talking about taking the functions that we do in-house -- development, business process management, business intelligence (BI) -- and moving those out into that cloud, where we can interact with the rest of the world, which is where companies make their money. You don't make money by selling your own bellybutton. You make money by going out and interacting with the rest of the world. That's where the opportunities are.

We're defining those right now, but it's what can I develop that I can push out to other people and what sort of process can I create to incorporate business partners? That's where the exciting stuff is, and we're really just wrapping a rope around that right now, and trying to tighten it up. There's a big loop right now, and we're still trying to squeeze value out of it.

Bifurcation in the market

Gardner: Do you see this bifurcation in the market, as I do, that developers grok this, and are into it, particularly the ones who are doing green-field and Web development, but that behind that partition, where operations and runtime environments are being closely monitored, they don't quite get this or not interested?

Meehan: Oh, absolutely. First, developers are always interested in doing new stuff. They're naturally curious that way. It's one of the good things about developers. They'll go out and try new things. Operations people are type A. They're not going to be into adding chaos to the mix.

That's always the natural tension you've got there. A lot of the PaaS vendors -- I'll mention one that's not on this call, but I know the folks at WaveMaker -- constantly talk about how you've got to have management tools.

That's being done to throw something back to those operations folks saying, "Okay, we can manage it. If we go in there, we use this platform, and it's not just pure chaos for you." Whether or not that's an essential part of the platform, there's an argument to be made that maybe it isn't, but it's being done as a sop to the people who are going to be naturally nervous about this.

Bryce: One thing I would add is that we've definitely seen with some of our enterprise customers is that development teams have come to us, because they are looking for that provisioning flexibility that we were talking about earlier. Their IT department can't get them the equipment that they need fast enough.

For example, last year we had a customer that had an application they were running that ended up needing a lot of capacity really quickly. In

The interesting thing is that once we've been able to build a relationship with a company's development department, we're starting to see IT departments embracing this

about a week, it went from running on a couple of servers to having 10x the amount of load and traffic that they were handling a week before. They said there was no way they could have gotten this many servers up and configured from the IT department.

The interesting thing is that once we've been able to build a relationship with a company's development department, we're starting to see IT departments embracing this, not as something to fear outside their walled garden, but actually a tool that's helpful for them.

There are a lot of things that we have to go through. We have to engage with them and, as Mike was saying, we have to help them feel comfortable about how the system is set up, how it's managed, what the process is all around, deployment, security and all of those kinds of things. Then, they can look at it as another resource, not something that's dangerous.

Gardner: So, comfort with resources. Let's go to Rourke McNamara at TIBCO. This is yet another element to integrate into the larger apparatus of the enterprise, something that the middleware vendors are designed to do. Do you at TIBCO view PaaS as an important new component that needs to be brought into a managed integration process of some kind?

Managed integration

McNamara: We absolutely do. Given that one of our strengths and a lot of our history is in integration software and large-scale platforms for developing distributed systems, TIBCO has a lot of experience with very large companies, and very large companies that develop a huge number of internal facing applications.

I'd totally agree that, up until this point, almost all of the work done in the cloud by enterprises has been Internet-facing, outwardly facing, or partner- or customer-facing applications. There's a huge interest in leveraging this technology and making use of the cloud for internal-facing applications.

And that said, the issue with the purely public cloud is that it's very complex and difficult to set up the security required to bring online an internal-facing application. You need a secure connection to your internal resources to integrate this with the rest of your infrastructure. You need a secure connection for your users to be able to access this. You need integration with directories and things like that, and all of that presents something of a barrier to entry.

We still know customers that have experimented with infrastructure as a service and PaaS for those types of applications, because they've heard all this hype. They've heard everything people said about cloud, and they want to make sure they understand what's going on. They're planning appropriately around these new developments. There's a lot of talk about some of the things that folks can use to mitigate those concerns.

"Private cloud" is a term that's been tossed around, but not clearly defined. There's a lot of confusion around it, because it's not clear.

. . . the issue with the purely public cloud is that it's very complex and difficult to set up the security required to bring online an internal-facing application.

Some folks just use it to mean a virtualization technology, and some use it more to refer to a full-on utility model for delivering compute power and computing resources.

Then there's the term that I think was first used about a year ago by a blogger -- "virtual private cloud" -- the notion that the resources exist outside your enterprise, but when you provision them, when you generate use for them, they are your resources. They're not connected to the public Internet. They're connected via some, sort of virtual private network connection to your internal network.

As you add to that collection of machines, those virtual resources are yours, and yours alone, fully secured, and effectively behind your firewall. Right now, it's just something folks are talking about. When that becomes a reality, it's going to be a huge influence in the way large enterprises look at cloud computing.

Gardner: Now, Rourke, wouldn't it be possible with this vision that you might create applications in the cloud and then you might want to host them in different places, depending upon how you would use that application.

Perhaps the application would be internal facing -- say, a business-to-employee (B2E) application, which you might put on one of those virtual private clouds. Or, you might have the same application that you use in the context of an ecology of business processes that we discussed with David Kelly.

You might want to have that in some respects in a cloud, but controlled with access, governance, and privilege policies and so forth. Then you might even want a very similar or a version of that application to be wide open, customer facing, the more the merrier. This all also sounds like a need for management policy governance and integration.

Exciting and scary

McNamara: Absolutely. That notion of a hybrid cloud, made up of all those different pieces and ones we haven't even thought of yet, is something that's both very exciting and very scary for large IT organizations.

It gives them an enormous range of flexibility, and it gives them an enormous range of power, but at the same time they're all worried about how they're going to be able to control. They worried about how they'll make sure that all of the stuff deployed in those environments is in compliance and talks together, make sure the right people can get into those different environments to change the appropriate applications and adjust the business logic, and that corporate standards are adhered too.

That's one of the things that we're looking very deeply at TIBCO Software. That's cloud governance, how we can expand the technology and experience we already have in service governance out to cloud governance, and to really take a look at that holistic hybrid cloud picture. We're trying to figure out how we can tie it all together for IT organizations, take away some of the complexity, and make it so that they can be very productive with those technologies.

Gardner: Mike Meehan, just as the world is becoming, apparently to some people, unsafe for service-oriented architecture (SOA), suddenly it's safer

Well, to knock a recent public fear on its backside, SOA is not dead.

for cloud governance. Are we talking about the same capabilities? Where companies may have been leery about horizontally applying SOA internally, they're going to have to resort to it anyway, because they are going to have to deal with these hybrid cloud issues.

Meehan: Well, to knock a recent public fear on its backside, SOA is not dead. All the things that we were doing inside of service orientation are still going on. I don't think you can move out to the cloud unless you service orient. One doesn't exist without the other. Of course, you're going to have to govern what comes in through the cloud, and that's a whole new layer of governance. We're seeing companies that are putting tools out there to help you do that.

Just recently, Vordel has an edge-networking device to help you govern your consumption of cloud-based resources. Appistry also has created some management tools along those lines.

We're seeing these companies taking what we were calling SOA management software and devices and pushing them out to the cloud. It's a fairly natural progression. TIBCO is absolutely right to be thinking in terms of this is the infrastructure you have to manage. You can't just suddenly grab things from outside and let them run wild internally. There has to be some adult supervision, and there's a huge business opportunity that comes with that.

As you use SaaS, as you bring in infrastructure and as you have your applications get infrastructure from cloud vendors, those are all going to be things that you're going to have to monitor and take care of as a business. You can't just say, "Well, that touches somebody else. Therefore, I can't care about it." This is your business. You have to care about it.

'Rogue' activities

Gardner: Let's take this to Jonathan Bryce. If you have developers that are perhaps doing rogue activities vis-à-vis your services, you would want to be a good partner with that enterprise in terms of its more official -- for lack of a better word -- IT activities. How do you bridge the gap, or do you leave that to somebody else?

Bryce: Developers come to us from larger companies. It's completely automated to get started. So, sometimes they come, and we don't even realize that they're using the system, until they're well into it.

Even these “rogue developers” are pretty diligent about making sure they inform us what they are doing. If they have special compliance requirements or governance needs, a lot of times they'll have questionnaires, security questionnaires, or operational questionnaires. We engage with different groups within our organization to complete those and help them get all that that information filled out.

Generally, the developers come here because they're trying to get some flexibility, speed, and capacity that they're not able to get internally. Most of them are fairly professional and straightforward about their organization. That's definitely what we prefer, because we want to build relationships with all of these enterprises over the long term, so that we can help them with more and more of their needs.

Gardner: What about the pace of this in the market? It seems to me that Amazon is moving very quickly. In fact, when they came up

Generally, the developers come here because they're trying to get some flexibility, speed, and capacity that they're not able to get internally.

with Amazon Web services, I don't think they anticipated that their test and dev portion of the business would become predominant. But that's okay, because where developers go the runtime usually follows.

We're starting to hear more about Java in this cloud. We're hearing more about .NET at Microsoft with Azure. We're hearing more about data crunching, vis-à-vis MapReduce-type technologies. What's your sense, Jonathan, given that you're a vendor in the business? Is Amazon in a sort of a mad dash to get out in front on this, and in a sense of dragging the other players along, and, as a result, they have to be there too?

Bryce: Amazon has created a great set of products. They've created infrastructure products that really are very similar to what people are already used to running. They also are kind of at a lower level. The lower level you go to, the more flexibility you have to customize it, and the more responsibility you have as well.

You can set up Amazon infrastructure and lock down network access. You can remove software from the system. You can have a lot of control over it. That's been a big part of their rapid growth. It's something that that's pretty similar to what people are already used to working with, but you can get it a lot faster and you can pay in a more utility manner.

Now, Microsoft is coming out with Azure, which is built around .NET. App Engine announced Java support. Our products have already been built around standard frameworks. To move that rapid adoption from infrastructure up to the higher levels, one of the keys is to try to work within the frameworks and the technology stacks that people are used to working with.

App Engine will see more rapid adoption of Java where they have implemented a lot of Java standards, the data API, the messaging API. Some of those things that they are adopting are official parts of the Java stack. I think that there will be more rapid adoption of that, versus the Python offering, which was a little bit nonstandard and a little bit out of the norm of what companies were used to dealing with.

This is what you see in every part of technology. Operating systems over time have standardized even desktop processing applications. You have a word processer, a spreadsheet, and presentation software. These things all standardize in technology, and that's when the adoption really spreads. That's what we're seeing right now as the PaaS offerings mature across the board.

Java in the cloud

Gardner: Now, Rourke McNamara, more Java in the cloud -- whether it's at Google or somewhere else -- means more enterprise-caliber, more transactional applications, and perhaps baseline business services coming across the cloud, all of which is probably good news for an integration vendor.

McNamara: Definitely. We've gotten a lot of interest from our customers in standards-based or normal frameworks and normal technologies in the cloud. The big barrier that we've seen with a lot of our customers to adopting PaaS has been that, up until very recently, just about none of them were standards based. They were a completely new and proprietary language or technology that was visual, something like Python with a custom API for Google App Engine or some sort of JavaScript for the other folks out there.

Our customers are scared enough of moving to this brand-new technology and taking that leap, but to do so in a way that they are building an application that can never run anywhere but in the cloud doesn't make sense for them. Dana, as you mentioned a couple of times, folks are more and more trying to build stuff that is reusable. If they build it for one particular application, they can re-purpose it for another application, or reuse parts of it for yet another application.

By using standards-based technologies -- Java or Spring, for example -- folks are able to build applications that are a bit more portable, a bit more reusable, and that aren't a giant leap into the cloud with something that they're not going to be able to bring back in house if the cloud thing doesn't work out for them.

Gardner: To this issue of cloud portability or cloud neutrality, we recently saw something called the Open Cloud Manifesto. I believe TIBCO was a signatory in support of that. Weren't they?

McNamara: No, we were not.

Gardner: You are not. How come?

McNamara: We got notice of that a little bit late in the game, and we just didn't have enough time to evaluate whether or not it makes sense.

By using standards-based technologies -- Java or Spring, for example -- folks are able to build applications that are a bit more portable, a bit more reusable

We didn't want to just rubber-stamp it, without really understanding what was going on, and there was a bit of a blowup there between, I think, Microsoft and Google over that.

Gardner: Right. But the general concepts of openness, neutrality, and portability, these are all important aspects from where you stand?

McNamara: They absolutely are, and as long as there is some sort of a manifesto, or some sort of set of standards and principles that are developed in an open fashion with everyone working together, we would definitely support that.

Gardner: Let's go around our group of analysts. Jim Kobielus, how do you feel about open standards? I guess it's like mom and apple pie, but how important is it for the total cloud ecology thing to develop and become a productivity and value to enterprise?

Kobielus: Very important. People often refer to "the cloud" by that phrase, as if this is a one unified platform, and it's the exact opposite. Every public cloud is essentially its own silo or stovepipe right now with its own standards and interfaces.

Gardner: Like the early Unix days, right?

Against the grain of SOA

Kobielus: Yup. As I pointed out in a recent article I wrote, right now the current state of cloud computing sort of goes against the grain of SOA. SOA is all about platform agnosticity and be able to port services flexibly and transparently from one operating platform to another, because they are all implementing these common SOA standards.

Then, when you talk about exporting them to the so-called cloud or the atmosphere full of clouds that's out there right now, the turbulent atmosphere, it becomes kind of absurd. Even a vendor like Microsoft who's both in the SOA world, and now increasingly with Azure in the cloud world, would have a disconnect between those two platforms.

They haven't really done a good job of closing the gap. If you've written applications to .NET and to their whole SOA platform, are they portable as is to Azure? No. Just by itself, a single vendor not getting its full SOA act together between the premises-based, and the cloud strategies should set off warning bells in every IT shop that has standardized with Microsoft, for example.

Gardner: It must be tempting though. If you could really lock up and create the "de-facto industry standard for cloud implementation and runtime, wow, that would be a really nice place to be. Just like it is if you can rack up the "de-facto standard" in on-premises runtime.

Kobielus: Right.

Gardner: Shall we expect them to try?

Kobielus: Of course they're going to try to dominate the cloud, just like they try to dominate the whole application server world, and they are not going to succeed.

Gardner: Jonathan Bryce, are you guys Baskin-Robbins when it comes to clouds -- 31 flavors, anybody's cloud will host it for you? What's your position on this openness?

Bryce: We mentioned earlier the Open Cloud Manifesto, which we were a signatory to. It was interesting the way that evolved. It was an idea among just a small group of people. It wasn't ever intended to really be, "This is the mandate." Then, a few of other companies started talking about it. The goal was really to put something out there that says there should be some sort of a real effort to standardize all these things.

Ironically, the process was not as open as it should have been, but the idea behind it is pretty valid. We definitely support the idea of standards, interoperability, and portability.

Rackspace has always been a supporter of various technologies. We always supported Windows, we always supported different distributions of Linux. We want to add value on top of the technology, versus being a technology vendor. So, we are not always going to be the cheapest option. We are not always going to be the commodity provider for something.

In terms of vendor lock-in, and requiring a specific deployment steps or development languages, that's not the stance that we take, because we would prefer to allow people to be able to run different applications in the environment that works best for them. Sometimes, that may not be us, but a lot of times we can add extra value on top of just the technology.

Part of the issue right now with all of these standards is that the providers and the vendors are all still developing our basic products lines. We're still developing what our niche is going to be and what our special sauce is? So, there hasn't been a lot time stick our heads up and say, "Okay, this is what they're doing, and this is what we're doing, and it makes sense for us to tie these together."

Picking one flavor

Gardner: Mike Meehan, do you think there is a temptation for enterprises to say, "We're a Java shop," or "We're a .NET shop? We only need to pick one flavor of cloud. As long as their services are loosely coupled, and we can interoperate, then maybe we're not too concerned about the underlying runtime. We'll look for the best synergy between the tools and the platform, the automation, and the plumbing." Does that make sense, and do you expect that we're going to see a rerun of the platform wars, but now at the cloud abstraction?

Meehan: Some shops will initially do what you just described. I don't think it makes sense. They'll do that because they haven't really thought it through. Openness is going to overrun that sort of thought. Hopefully, enough users have figured out that you need to be so far beyond Java vs .NET at this point in your life that this really shouldn't be the discussion that’s front of mind anymore.

Obviously, there are programmers who know one versus the other, but it's not what's driving your business, and there are larger concerns. The cloud allows you to get to different layers of abstraction and larger concerns, and you shouldn't just necessarily be Java and .NET.

So, this whole notion that somebody has to roll this whole thing up is wrong -- I'm just going to go right to Microsoft and get it from them. I

What we've seen from Microsoft to date hasn’t made a lot of sense. That's why Google hasn't really taken off yet, because it hasn't done something quite elastic enough yet.

agree with Jim. If they try that, they are going to fail, and they're going to fail hard. I hope they have a sense of that, and I hope they're looking at this as a more inclusive universe, rather than a more exclusive universe. Amazon is doing really well right now with EC2, because it is most generic. For example, Sun can put Rails 2 on top of Solaris, and put that up in EC2.

Gardner: We also have Red Hat Enterprise Linux available is a runtime environment?

Meehan: Exactly, and that's all just in there. You can throw anything in there. It really is elastic. Elasticity is what's got traction for Amazon right now. It's not that Amazon has the most unified, elegant platform offering out there. What we've seen from Microsoft to date hasn’t made a lot of sense. That's why Google hasn't really taken off yet, because it hasn't done something quite elastic enough yet.

Gardner: We haven't mentioned, but they came from the SaaS side, and they have an interesting balance between proprietary and open. They've tried to create an ecology, but they also certainly want to be a destination that makes more sense to stay than go. “How do you view what Salesforce has done, and is that the model that we will see -- that "cloud of clouds" model.

Meehan: What's interesting about Salesforce is that, as you said, they just started doing stuff, and it gained traction. Now, you're thinking of them as platform. You're not going to go pick and grade various pieces of SaaS.

Gardner: More than just one application as a service.

Meehan: Exactly. There's something to just getting out there and doing stuff. I don't think they've bottled the way they to do it, but they have come up with a nifty process for getting there, which is, get involved, start getting stuff out there, and you'll find out what sticks. We're all theorizing about what's going to happen? Salesforce is an example. What's going to happen is going to be a function of what we do, and we are just not going to know until we do it. So let's start getting working.

The pain of transition

Gardner: Dave Kelly, it's a little easier for someone like Amazon that's got a retail business to draw on and to monetize on its investment for infrastructure. It's another thing for Google that come in with its advertising base to create these services. They can make these services in the business models that they have dovetail, with this new cloud model.

For somebody like Microsoft, coming from an older licensed software business and trying to re-orient and re-factor that into a pay-as-you-go service, doesn't that involve some significant pain making that transition?

Kelly: It should. This a real challenge for Microsoft. It's like the open systems discussion we had a little while ago. It makes more sense for players that actually earn their revenue in a different form than traditional operators, because someone like Amazon has a core business.

Someone like Microsoft is kind of painted into the corner at the moment. That's a challenge not just for Microsoft, but for other traditional vendors.

They can expand into this new area by offering low-cost services that take away from competitors, but don't hurt their core business.

Someone like Microsoft is kind of painted into the corner at the moment. That's a challenge not just for Microsoft, but for other traditional vendors. I think there are opportunities. With Salesforce, you can make that same kind of argument. That's more an environment for that sort of solution specific type scenarios related to Salesforce. I don't see it growing into a broader, more common application platform. It could, but I don't see that.

Gardner: Okay, we're just about out of time. Let's go to Jonathan. Is that what we were seeing now -- a bunch of barbarians at the gate in the form Salesforce, Google, and perhaps Mosso as well?

Bryce: Perhaps. Maybe it's something friendlier than barbarians, but there's one thing that's important to add to the discussion of openness, standards, and the point about Amazon being very flexible and something that you can adapt in a lot of ways. We aren't seeing the potential of cloud yet. We're going to see the real potential of cloud, when there's some interoperability and portability that allows just amazing applications and systems of systems to be built on top of all of these clouds.

That's when we're going to see how this shift in computing is going to change the world. When you can develop against multiple clouds, when you can get storage in multiple occasions, when you can get as much compute as you need to run a simulation or a model, and have thousand of servers, and all of these kinds of things together, that's when it's really going to be amazing.

To do that, you will see vendors building businesses on top of the cloud. We're not going to build every feature that's going to be implemented on our infrastructure. There are ecosystems that have developed around Amazon, us, and the other providers. And we know that ecosystems will start to join together, and we'll come up with some really great tools, some really great new technologies, that harness all this power. That's when this is going to be a very exciting space.

Gardner: Okay, Rourke, last comment to you. How do you view these cloud, permutations, particularly PaaS? Are they barbarians at the gate or liberators?

PaaS as 'liberator'

McNamara: I definitely view PaaS and some of these other permutations of cloud service as liberators. If you look at what PaaS really does for developers and IT organizations, it frees them from having to worry about a bunch of details that have nothing to do with their core business and don't even have anything to do with the application that they are writing.

It frees them from having to install platform software on a bunch of machines, putting those machines into racks, connecting them up to the management and monitoring infrastructure, and getting everything set-up so that those machines are fault-tolerant and the loads distributing appropriately. It frees them from making sure that they have the right machines to handle the load, and making sure that they are predicting load increases and capacity increase or requirement increases, and far enough in advance that they're able to buy new machines.

All that stuff has nothing to do with delivering new functionality to their business users. That's really what their job is, delivering new functionality to their business users. PaaS lets them focus ideally on delivering new functionality, and lets someone else worry about all of those of details.

Gardner: Picks and shovels, to the liberators, right?

McNamara: Exactly.

Kobielus: I second that. It's liberation, because, to carry forward the scenario that I described a little while ago, you should be able to develop your analytical data mart and test it inside an Amazon cloud. But, if the Google cloud gives you a cheaper hosting, then you should be able to flexibly migrate that entire analytical data mart after production to the Google cloud where it is hosted for X number of months and years, until you're ready to deploy your premises-based data warehousing cloud, a petabyte-scale cloud.

It should be migrated from Google to your premises, and back and forth, depending on your changing operational and performance needs. For this kind of a project, a data mart for data mining, the PaaS model should give you that flexibility and freedom of not having to worry excessively about the hosting issues.

Gardner: Well, we'll have to leave it there for today. Please allow me to thank our guests. Jim Kobielus, senior analyst at Forrester. I appreciate your time, Jim.

Kobielus: Sure.

Gardner: David A Kelly, president of Upside Research. I appreciate it, Dave.

Kelly: Always fun to be here.

Gardner: Mike Meehan, senior analyst at Current Analysis. Always a pleasure Mike.

Meehan: Thank you, Dana. I just want to add that Paas, is, as far as I know, a maker of Easter egg color dyes.

Gardner: It's rubbing off this time of year. There's another pun for you.

Bryce: It's full of PaaS-abilities.

Gardner: Jonathan, PaaS is not passé either. Jonathan Bryce, co-founder of Mosso at Rackspace. Thanks for joining us.

Bryce: Thanks for having me.

Gardner: And, Rourke McNamara, product marketing director at TIBCO Software. Thank you, Rourke.

McNamara: Thank you, Dana.

Gardner: I also want to thank our charter sponsor for the BriefingsDirect Analyst Insights Edition Podcast series, Active Endpoints, maker of the ActiveVos visual orchestration system, as well as continued underwriting from TIBCO Software.

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 and Charter Sponsor: Active Endpoints. Sponsor: TIBCO Software.

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

Edited transcript of BriefingsDirect Analyst Insights Edition podcast, Vol. 40 on the promises of platform as a service and its apparent lack of traction among enterprise managers. Copyright Interarbor Solutions, LLC, 2005-2009. All rights reserved.

No comments:

Post a Comment