Transcript of a BriefingsDirect podcast on cloud adoption best practices with HP executives.
Listen to the podcast. Download the podcast. Find it on iTunes/iPod. Learn more. 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 cloud computing, and how enterprises can best prepare to take advantage of this shift in IT resources use and acquisition -- but while also avoiding risks and uncertainty.
Much has been said about cloud computing in 2008, and still many knowledgeable IT people scratch their heads over what it really means. We’ll dig into the hype and opportunity for cloud computing with executives from Hewlett-Packard (HP) and EDS, an HP company. We'll discuss the pragmatic benefits -- and also the limits and areas of lingering immaturity for cloud-based delivery of mission-critical applications and data.
Here to provide the inside story on the current state of cloud computing we welcome our panel, Rebecca Lawson, Director of Service Management and Cloud Solutions at HP. Welcome to the show, Rebecca.
Rebecca Lawson: Thank you.
Gardner: Next, Scott McClellan, Vice President and Chief Technologist of Scalable Computing and Infrastructure in HP’s Technology Solutions Group (TSG). Welcome, Scott.
Scott McClellan: Thank you.
Gardner: And last, Norman Lindsey, Chief Architect for Flexible Computing Services at EDS, an HP company. Welcome, Norman.
Norman Lindsey: Thank you, sir.
Gardner: The trends and the talk around cloud have jumped around a fairly large landscape -- everything from social networking computing to Web services, video, and ... you name it. But what we are going to be talking about is primarily of interest to enterprises, and what we could continue to classify as utility or grid-type computing. First, I want to talk to Rebecca about what is changing around cloud computing, and why IT people should be taking this seriously at this time.
Lawson: Let me first say that at HP, we are really interested in just trying to articulate where we see cloud opportunities -- and how they differ from existing infrastructure, application and service environments. So the way that we define cloud at HP is that we consider it a means by which very particular types of highly scalable and elastic services can be consumed over the Internet through a low-touch, pay-per-use business model.
There is an implication with cloud that is different. It solves different problems than what we have been solving over the last few years, and it implicates both breakthroughs in technology architecture and the confluence of that with new business models.
That’s kind of a mouthful, but we basically think that the enterprise should be aware of what’s happening at the infrastructure level, at the platform level, and at the application level -- and understand what opportunities they have to further source from the cloud certain services that will directly relate to the business outcomes that their organizations are trying to achieve.
Gardner: Do you think the interest at this time is primarily an economic story, or is it convenience? What are the drivers behind all this interest in cloud computing?
Lawson: There is an overriding notion that the cloud provides a lower-cost option for computing, and that may be true in a very few limited use cases. Really, from an enterprise point of view, when they are running mission-critical applications that need security and reliability, and are operating with service-level agreements (SLAs), etc., the cloud isn’t quite ready for prime time yet. There are both technical and business reasons why that’s the case.
As far as the idea of the cost savings, it’s good to look at why that is the case in a few certain areas, and then to think about how you can reduce the cost in your own infrastructure by using automation and virtualization technologies that are available today, and that are also used in the “cloud.” But, that doesn’t mean you have to go out to the cloud to automate and virtualize to reduce some cost in your infrastructure.
Gardner: Let’s go to Norm Lindsey. There are a number of other similar overlapping trends afoot today. There’s virtualization at a number of different levels, application modernization, consolidation, next-generation data center architectures, services-oriented architecture (SOA), an emphasis on IT service management.
Does cloud intersect with these? Is cloud a result of some of these? What is, in a sense, the relationship between some of these technology trends and these economics-driven cloud initiatives?
Lindsey: A lot of these technologies are enablers for a cloud approach to services. The cloud is an evolution of other ideas that have come before it, grid, and before that Web services. All these things combine to enable people to start thinking of this as delivering service with a different business model, where we are paying for it by the unit, or in advance, or after the fact.
Virtualization and these other approaches enable the cloud, but they aren’t necessarily the cloud. What IT departments have to do is start to think about what is it they’re trying to accomplish, what business problem they’re trying to address, as they look at cloud providers or cloud technologies to try and help solve those problems.
Gardner: It also seems that we are hearing about private clouds or on-premises use of these architectural approaches, as well as public clouds or third-party sourcing for either applications or infrastructure resources. Does this boil down to a service orientation, regardless of the sourcing? Perhaps you could help people better understand the different between a private cloud and a public cloud?
Lindsey: Private cloud versus public cloud is part of this whole evolution that we’ve seen. We’ve seen people do their own private utilities versus public utilities such as flexible computing services provide. The idea of a private utility is that, within an organization, they agree to share resources and allow the boundaries to slide back and forth to hit the best utilization out of the fixed set of assets or maybe a growing set of assets.
Nevertheless, they agree to share it to try and approve the utilization. The same idea is in a public utility or a public cloud, except that now a third party is providing those assets and providing that as a service. It increases the concerns and considerations that you have to bring to the party. You have to think about problems that you didn’t have to think about when you had a private utility.
When you go to a public space, security is paramount. What do I do with my proprietary information and service levels? How certain can I get what I need when I need it. The promise with the cloud is great, but the uncertainty has caused people to come up short and decide maybe it’s better if I do it myself, versus utilizing an outside service.
Gardner: Now, I think it’s fair to say that, at this point, this is all still quite new and experimental -- with developers, small companies, and some departments -- using such resources as Amazon Web Services. Clearly this is still in the very early innings, but some of the analyst firms are predicting as much as 5 percent of IT might be devoted to this in several years. While that’s a fairly large number in total, it’s still quite small in regard to the whole pie.
Let’s go to Scott McClellan. Are there really serious positive business outcomes that should entice organizations to start looking at cloud computing now?
McClellan: I definitely think there are. Basically I see the conversation happening between business and IT in two different ways, and one of them was already touched on earlier, when you were talking to Rebecca.
That has to do with the cost factor. That’s your business asking your IT department to reduce cost; CEOs put pressure on CIOs to deliver more with less.
So there are aspects of automation and virtualization that allow you to get to a more utilitized approach to delivering the services within your IT department -- to allow you to increase flexibility, reduce cost, drive up utilization, and things like that to address the cost issue. So there are real business drivers behind that, and that’s especially heightened in today’s economic climate.
In the longer term, the more overarching impact of cloud comes when your IT department can deliver value back to the business, rather than just taking cost out. Some examples of that are using aspects of social networking and other aspects of cloud computing, and the fact that cloud is delivered over ubiquitous media, the Internet, to increase share of wallet, increase market share, maybe bring higher margin to a business, and build ecosystems, and drive user communities for a business. That’s where cloud brings value to a business and that’s obviously important.
Gardner: So we have, at one level, an opportunity to take advantage of these technologies for pure efficiency’s sake for our internal IT operations. There is also this additional opportunity to use the clouds as a gateway to new or existing customers and be able to service them perhaps better through this ubiquitous medium of the Internet and perhaps at lower cost. Is that right?
McClellan: Yeah, it’s absolutely true. The former, the taking cost out is the first way. The first wave of innovation from cloud computing is coming from making services consumable on a different model, on more of a utilitized model, and that drives up utilization, etc. To unlock some of the value requires innovating at the application tier, in many cases, but absolutely you can bring both benefits to a business.
Lawson: I’ll give a concrete example of this cost. Let’s choose an example, first of a service your business needs to have -- a credit check service. Obviously, when you are selling a product, you want to make sure that your customer has credit, which, of course, is all the rage today.
You could think of a credit-check service as having a very specific business outcome. It may be that your company has an internally developed service that maybe you built, and it’s tied into your SAP, Ariba, or what have you.
Or, it may be that your credit-check service is hosted by an external service provider, but still designed in a traditional architectural manner. Or, it may be that there are credit-check services available through the cloud, designed in a different application architectural style that suits your purpose.
Either way, what IT is going to need to do is really think through its service centric way of behaving and a way of operating IT -- so that what’s appropriate for that company can be arbitrated by IT, knowing that they have to take into consideration security, speed, and accuracy. So for some companies, doing a credit check through a cloud service might be perfectly fine. For other companies, it may be way too risky for them for whatever reason.
We need to think in terms of which services provide what level of value, based on the complexion of that particular company -- and it’s never going to be the same for all companies. Some companies can use Google Gmail as an email service. Other companies wouldn’t touch it with a 10-foot pole, maybe for reasons of security, data integrity, access rights, regulations, or what have you. So weighing the value is going to become the critical thing for IT.
Gardner: It appears that the ability to take advantage of cloud computing comes from an increased services orientation, and understanding the technologies and how to take advantage of them and exploit them -- but that the larger business decisions really are around which services should or shouldn’t be sourced in a certain way, and what level of comfort and risk aversion are acceptable.
This is probably going to be something that needs to be judged and managed company-by-company, even department-by-department.
How do companies start to get a handle around that decision process which seems critical -- not just how to take advantage of the technology but in which fashion should these services be acquired and managed?
Let’s go to Norm. How do people start managing, at a local individual level, the decision process around which services might become cloud services?
Lindsey: Start by looking at the business problem that you are trying to solve, and IT has to start looking at the requirements and dealing with it as a requirements issue, as opposed to a technical issue. They need to make sure that the requirements are clear and all stakeholders understand what you are doing.
Then you can start to look around at your internal capabilities, versus external, and make some decisions as to how you want to solve that problem, whether buying an external service or creating a service internally and delivering it to your customers with your own internal utility.
Gardner: Rebecca, this raises the question, then, of … Who owns this decision-making process around cloud, utilization, and/or resource? This seems to be an abstraction above IT, but you certainly need to know what IT processes are involved here.
I know we are early in this, but is there any sense of how who owns the decision-making process around cloud is going to shake out?
Lawson: That’s a really great question, because a lot of people in the lines of business or business functions can go out to the Internet and make a decision. “Hey! We’re going to use Salesforce.com,” or what have you. Those decisions made without IT could have some really deep ripple effects that a line-of-business person might not realize.
People in the lines of business don’t think about data architecture and integrity, they don’t think about firewalls, they don’t think about disaster recovery, and they shouldn’t. That’s not their job.
So this will force IT to come closer to the people in the business and really understand what is the business objective, and then find the right service that maps to the value of that objective. Again, we can’t emphasize it enough. This should really change behavioral dynamics in IT and how they think about what their job is.
Lindsey: That’s a key point -- the IT guys become an enabler, as opposed to a gatekeeper. They know what the compliance issues are; they know what the regulatory rules are on their company to meet Sarbanes-Oxley, or whatever world they live in.
The line of business has the business problem and they need to focus on what their problem is and let IT answer the question in terms of, “These are some possible solutions. This is what they cost. Now tell me which one you do.” But these will all have to meet the myriad list of requirements that we have to live within.
Gardner: It appears to me that there are a couple of different levels of risk here. One risk would be that people start jumping into cloud and external-service consumption piecemeal, without it being governed or managed centrally, or with some level of oversight in a holistic sense.
The other risk might be that you are so clamped down, and you are so centralized and tightly managed, that no one takes advantage of efficiencies that become available through the cloud. You then have unfortunate costs and an inability to adapt quickly.
Let’s go to Scott McClellan. How are companies expected to manage these types of risks, that is to say, over-consumption or under-consumption of cloud services? How can companies become more rational in how they approach these issues?
McClellan: In the process of getting to a service-centric IT governance model, they’re going to have to deal with the governance model for deploying new services. Again, I think risk is partly a function of benefit. So when there is a marginal benefit or when the stakes are very high, you would want to be very conservative in terms of your risk profile.
Basically, within the spectrum of things that are cloud computing, you have everything from infrastructure as a service … all the way up through virtualized infrastructure, a platform on top of that, an application on top of that, or perhaps a completely re-architected true cloud-computing offering.
As you move up that spectrum, I think the benefits increase, but in not all cases are the application domains available in all of those environments.
There are several choice points here. What services are available through some cloud model, what model of availability, what are the characteristics of that model, what are the requirements for that particular service – and what are the security performance, continuity integration, and compliance requirements? Those all have to be taken in holistically and through a governance model to make the decision whether we are going to move from the traditional deployment model to a cloud-delivery model, and if so, which one.
Gardner: To me, this governance issue sounds an awful lot like what we’ve heard around SOA, and what you need to put in place to take advantage of that approach.
Rebecca, are we talking really about the same set of issues that, if you put in a good SOA infrastructure, management, governance, and capability set -- and if you organize your culture and your people to think about services – that that puts you in a good position to manage cloud? You can find were it’s appropriate, and then be able to find that balance between these risks?
Lawson: That’s a good observation, and there is a parallel between the notions of SOA, the loose coupling of services, and what we’re talking about here. The hard part is that services come in many different flavors and architectural styles. So in reality you might be managing a service that runs on a very old architectural style, but it really delivers value. You really want to maintain it, and it’s worth it. You might also want to adopt a Web-oriented architectural approach, vis-à-vis using some cloud services in another part of the organization.
The parallel is there. People who’ve grown up through a SOA kind of model naturally gravitate to this. The service provider and consumer relationship is a big change with cloud because, all of a sudden, providers look different than they used to.
Companies that you didn’t think of as service providers are now a service provider. You never used to think of Amazon as a company you might go to to get compute from. You used to buy books there.
So what happened? All of a sudden, lots of people can become providers in startling ways, which is great. It’s a whole new burst of creativity and possibility in the area of technology-enabled services. Obviously, we have to tread carefully, because businesses have to grow, and you’ve got to choose wisely.
Gardner: I wonder if there are other precursors to organizations being better able to take advantage of cloud computing, but at low risk. I suppose one would be IT service management, treating IT as a bureau or service provider, the charge back type of system.
Any input, Norm, on some of these other precursors that organizations might think about as they start to wonder how they can best take advantage of cloud?
Lindsey: Actually, one of them is one you haven’t brought up, which is a lot of times they are out of space and out of time. They have some idea or they have some new business. They want to load it and they are out of room in their data center.
Or it’s something that just comes up really quickly, and they need to act quickly. The flexibility and the nimbleness of the cloud enable them to respond. So, as far as the drivers inside the business, that’s one of the big ones. The other one is just running out of power and space inside of their existing facility.
Gardner: I suppose that gives them the opportunity to ramp up, but without a whole lot of upfront capital expense. They can pay for this on a per-use basis, right?
Lindsey: Precisely. You rent instead of buying. The other obvious benefit is that you have minimized your risk and you can turn it off, if things don’t go the way you want them to.
Gardner: Let’s look at some of the things that cloud computing can’t do so well. Obviously, as they say, we are in the early innings here. Let’s go to Scott McClellan on this. Not all applications can be delivered by a cloud. There are design and data issues and application programming interface (API) issues. We’re not ready for database joins and two-phase commits, and needs around transactional integrity where you need to have correction of transactions, and so forth.
Maybe you can help our listeners understand, at least for the foreseeable future, what types of applications and services might be appropriate for cloud -- and which ones would not be?
McClellan: It’s partly a matter of how modern is the application architecture that enables the service. So, it is a bit of a continuum. To some extent, the question isn’t, “Can it be delivered as a service model?” but “Can it be delivered in as a service model at the necessary scale on a cost curve that allows the service to be delivered at an attractive price?”
So it’s not a simple black and white. Is it possible to do this particular service in the cloud? You might be able to take a legacy architected application, delivered it in, say, software-as-a-service (SaaS) model, assuming it’s basic underlying architecture is relatively modern, and it can be Web-enabled and it has appropriate user interfaces and so forth to be Web-enabled.
The immaturity of some of the data services and the truly scalable cloud computing infrastructure -- examples are things like Google’s BigTable or Hadoop data-services level -- do provide some relational data semantics, but they are nowhere near as rich as the full database semantics provided by the mature database management subsystems. As you mentioned there is no way to do a join.
Gardner: It seems an important hurdle to overcome in taking advantage of cloud would be the proper mixing, if you will, of data. There needs to be some kind of a sharing, where not the entire database, but perhaps a level of meta data might be shared between different organizations, private and public.
Do you have any thoughts, Rebecca, on how HP views that sharing, that data issue? Again, that’s something for an IT department, or may be even a marketing department, to tackle.
Lawson: Obviously, there will be data that you just don’t want to share with anyone, but there is a good use-case out in the cloud for a provider to offer up a ton of data that might be valuable to a whole bunch of different consumers. Let’s say it’s demographic data, and they may want to make a marketer’s ability to access that data through a number of services very agile and very scalable. That would be an example of a potential place where somebody could write some cloud-based services or applications and offer them through the cloud.
Intelligence in data varies widely, so it’s hard to generalize. On the other extreme, inside the firewall, you might have some extremely rigorous requirements for what data goes into your enterprise data warehouse, who gets to access it, how the tables are set up, or what the security provisions are. That would be another extreme where you have no interest whatsoever in sharing that with anyone, and it’s considered core to the company.
So that’s a great example of where you have to really consider the value of the service and the output. What’s the business outcome and how should we think about where we let our data live, how we access our data, how we mash it up with other information sources. Again, the bad news is there is no simple answer; the good news is there are lots of opportunities to get very clear in what you want as a result of that data, and lots of places to get it.
Gardner: All right, let's give the last word to Scott. Clearly, the technologies are there for a scalable and agile infrastructure. The economics are apparently quite compelling.
This comes back down then to the organization behavioral risk management issue. My last question to you is, in a period of economic downturn where economics and cost issues are paramount, is cloud computing something that will be accelerated by the tough economic times, or will people back off from something like what cloud offers until they have a better picture in terms of growth?
McClellan: My personal prediction would be that the tougher economic conditions would heighten the acceleration of cloud computing, and not just because of the opportunity to save cost. Reinforcing what we brought up earlier, there are some clear opportunities to bring value to your business.
Examples of that are things like being able to drive user communities, users and consumers of whatever it is your business produces, using techniques of social networking, and things like that.
There is the question of how to use the advantages you get from cloud computing to drive differentiation for your business versus your competitors, because they’re hesitating, or not using it, because they’re being risk-averse. In addition, that compliments the benefits you get from cost savings.
The other characteristic that the tough economic conditions could have on adoption of cloud computing is that it might cause customers to shy away from particularly painful places, where the risk is super-high, but it will kind of lower the barrier or the threshold that you have to clear for the opportunities that are less extremely risky, if that makes sense.
Gardner: I think you are talking about the high upfront capital outlays to start something. If you build it, you hope they will come, that kind of thing?
McClellan: That's on the service-provider side. There could be some risk aversion on service providers building out giant infrastructures, with just the hope that someone will come and consume them. I agree with your point there.
What I really meant is that, if you are an IT shop and you are trying to decide what to move to a cloud paradigm or a cloud model, you’re likely to really focus on the places where either you can get that big win -- because moving this particular service to a cloud paradigm is going to bring you some positive differentiation, some value to your company.
Or, you are going to get that big cost savings from the places where it's the most mission-critical -- the place where you have the least tolerance for downtime, and you have the greatest continuity requirements, or where the performance SLA has been most stringent. The thinking may be, “Well, we’ll tackle that later. We’re not going to take a risk on something like that right now.”
In the places where the risk is not as great -- and the reward either in terms of cost or value looks good -- the current economic conditions are just going to accelerate the adoption of cloud computing in enterprises for those areas. And they definitely do exist.
Gardner: It gives companies a series of additional choices at a time when that might be exactly what they need.
McClellan: That's right. And in some cases, it's not super-expensive to move to this model, and you'll have a quick payback in terms of return on investment (ROI). If you are bringing value to your company and differentiation, this is a good time to do that. Strike while there is a sense of urgency. It creates a sense of urgency to strike. I guess I would say it that way.
Gardner: We’ve been discussing some of the advantages and potential pitfalls of cloud computing. It seems that the opportunities are there for those who examine it carefully and appropriately, and can balance the risks to get the rewards.
We’ve been chatting today with Rebecca Lawson, the Director of Service Management and Cloud Solutions at HP. Thanks, Rebecca.
Lawson: Thank you.
Gardner: Also, Scott McClellan, Vice President and Chief Technologist of Scalable Computing and Infrastructure at HP's Technology Solutions Group. Thanks so much, Scott.
McClellan: Thank you very much. I appreciate the opportunity.
Gardner: And also, Norman Lindsey, Chief Architect for Flexible Computing Services at EDS.
This is Dana Gardner, principal analyst at Interarbor Solutions. You 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. Sponsor: Hewlett-Packard.
For more information on HP Adaptive Infrastructure, go to:
Transcript of a BriefingsDirect podcast on cloud adoption best practices with HP and EDS executives. Copyright Interarbor Solutions, LLC, 2005-2008. All rights reserved.