Listen to the podcast. Find it on iTunes/iPod and Podcast.com. Download the transcript. Sponsor: The Open Group.
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 in conjunction with The Open Group’s
Enterprise Architecture Practitioners Conference held in Seattle, the week of Feb. 1, 2010.We've assembled a panel to examine service-oriented architecture (SOA) and cloud computing -- the relationships, the inter-reliance and the realities. Three years ago, the IT transformation poster child was SOA, and now we're well into the hype curve around cloud computing, but has one actually given way to the other? Are they linear in their relationship, or perhaps mutually dependent in some ways, and to what degree?
We’ll explore today whether SOA has found new value and relevance as a foundation and perhaps catalyst for cloud computing, especially for so-called private clouds. And, we'll see how the emergence of SOA and cloud may be happening in different places inside of enterprises. Shouldn’t one hand get to quickly know what the other is up to and perhaps even work together?
There are a series of podcasts from The Open Group conference: on cloud computing, enterprise architecture, business architecture, Archimate, and cloud security.
Here with us now, however, to plumb the depths of how SOA and cloud computing do or don’t come together, are a SOA expert from the Open Group, as well as an Oracle expert on architecture, and a Salesforce.com executive on cloud computing.
We're here with Dr. Chris Harding, director of the SOA Work Group at The Open Group. Welcome to the show, Chris.
Dr. Chris Harding: Thank you, Dana.
Gardner: We're also here with Stephen G. Bennett, Senior Enterprise Architect at Oracle. Welcome, Stephen.
Stephen G. Bennett: Thank you. Happy to be here.
Gardner: And, we're here with Peter Coffee, Director of Platform Search for Salesforce.com. Hello, Peter.
Peter Coffee: Hi. Good to be with you all.
Gardner: Let me take our first question to you, Chris. You’ve been working with SOA for some time with a work group at The Open Group. Has the SOA conversation changed in the last several years due to the interest in cloud computing?
Harding: Well, Dana, the conversation has changed in that, five years ago, when we started
getting into SOA, there was a huge amount of excitement and a great deal of buzz about it. Now, we can see that the hype cycle has run its course, but we're still seeing a great deal of technical interest in SOA and we're also seeing that companies are using it and are increasing their use of it. So, there is a steady uptake in the use of SOA, although the excitement about it has died down.Gardner: That’s an interesting effect, where the interest seems to be less, but the actual implementation is more. Peter Coffee, do you have any sense of how SOA is working or not working in your efforts around cloud computing?
Considerable upside
Coffee: I've been covering SOA for a long time. Before I was at salesforce.com, I was with the
industry journal eWEEK, so I had a lot of opportunity to talk with people who were in the evaluation and adoption phase. I'd say the people who adopted SOA in the previous decade got considerable upside, but those who did not didn’t really suffer any penalty for not doing so.In the situation we're in now, where the economics of cloud computing are becoming quite compelling, the downside of not having a SOA is becoming quite apparent. If you don’t have a service environment, then your ability to extend your current assets and integrate them with cloud services is going to be somewhat hampered.
So, people are realizing now that the wait-and-see option is more perilous than it used to be. This is accelerating the actual adoption of what we would call SOA, except that’s no longer the label du jour.
Gardner: Stephen Bennett, from an architecture perspective, do you also see that the SOA activities are no longer a nice-to-have, particularly if you're interested in working towards grid, cloud, utility types of computing models?
Bennett: It's definitely a necessity, and the majority of large enterprises today are doing SOA in
one fashion or another at different levels of maturity, whether that’s from the quite immature approach of seeing it as a pure integration play all the way up to seeing it more as a business agility kind of play.So, it's becoming a norm and, therefore, we don’t need to keep hyping it or pushing it. We need to use the characteristics it offers with other supporting technology strategies such as cloud.
Gardner: Back to you, Chris Harding. When we think about how to approach SOA and about some of the business models that seem to be attractive around cloud computing, do you think that they're happening in different places. Is there a dissonance of some sort in some of the enterprises? That's to say is one hand doing SOA, the other doing cloud, and they're not necessarily coming together very much.
Harding: Enterprises are complex organisms, and often different parts of them aren't completely in sync. It’s very interesting that service orientation is very much a business concept, and SOA has been about the application of that business concept to the technology. Cloud computing, on the other hand, is very much a technical concept. It’s about what you can do with technology over the Internet.
It is a technical concept, but it has had really a big impact on the business structure. So you can see them as complementary. SOA has been the application of business principles into the technology. Cloud is a technical concept, which has had a huge impact on the business. So, yes, there probably are different parts of the organizations looking at cloud and looking at SOA, but there is a big dynamic that says they should be working together on both of them.
Gardner: How about from your perspective, Peter Coffee? Do you see different cultures, priorities, or efforts at work that haven’t quite matched up or connected yet?
Two perspectives
Coffee: I certainly do, and I’d like to look at that from two perspectives. First, without actually
disagreeing with Chris, I see things slightly differently. It seems to me that SOA very quickly became a label of products that vendors wanted to sell. So, you saw a lot of things like enterprise service bus (ESB) products and so on.It became dangerously easy to think that you were doing SOA, if you were buying the tools and failing to appreciate how much of a cultural and management achievement it was to get people to think of themselves not as owners of and the gatekeepers to an IT asset, but instead being publishers of and supporters of a service to other parts of the business.
The second point goes back to what was said just a moment ago by Steven. It’s absolutely critical to understand that you can view SOA as simply a way of integrating the stuff you have, or you can move to the next level and start to think of it as the way you do your business. The way your business units interact with and support each other with the technology is just the enabler for that.
The same is true of the cloud. It's possible to take the existing IT model of isolated applications, each with their own data stores, and replicate that model in the cloud with elastic scalability of capacity. That would be the level of the cloud industry that’s typically called infrastructure as a service (IaaS).
Or, it's possible to use the cloud as a much more interesting and fluid medium for interaction among much more granular and business-oriented services at the level that’s traditionally been called in the industry either platform as a service (PaaS) or software as a service (SaaS). It depends on the level at which you choose to consume other people’s application work, instead of doing new application development yourself.
So, it’s really important to realize the orthogonality between the technology you're adopting and the manner in which you are using it. It’s possible to do SOA without the cloud. It’s possible to do better SOA with it. It is also possible to do an isolated silo-oriented architecture locally and also to do that in a cloud environment. Neither one necessarily implies or impels the other.
Gardner: But, they certainly seem to have a mutual reinforcing benefit for one another.
It's a combination of technologies that are finally ready for prime time, and an ecosystem that’s ready to support those technologies well.
Coffee: Exactly. As I said earlier, the economics of being able to have elastically scalable capacity to be able to handle peak loads without needing to own the peak capacity and wind up with very low utilization rates on your capacity are becoming so compelling that people are asking how they're going to take advantage of this opportunity of this cloud environment.
It's a combination of technologies that are finally ready for prime time, and an ecosystem that’s ready to support those technologies well -- providers of services and providers of expert assistance in using those services.
That’s a very important enabling ware, when your major system integration firms begin fully to understand how they can incorporate cloud services into the portfolio of technologies that they make available to their customers. When you put that all together, the downside of not moving to an SOA becomes an embarrassing lack of ability to take advantages of these incredible economies.
Gardner: Stephen Bennett, we've been working through maturity models and technology and adoption patterns, but we also had a major recession globally, perhaps the worst in 75 years. One of the things that many people thought derailed the adoption of SOA was the lack of funding for added-on expense or discretionary IT spending.
At the same time, we had cloud come in and say, "We can do more for less, increase your productivity, and perhaps add more capacity into your data centers, preventing you from having to invest in additional ones." It seems that cloud has given SOA an economic boost in terms of the rationale for investing in SOA. Does that make sense to you?
Pragmatic and realistic
Bennett: It definitely makes sense, but we also have to be pragmatic and realistic here. Any disruptive approach or technology such as cloud and SOA -- backing up what Peter said -- is not a pure product or technology play. There are other considerations. You have to take the people and process into account. That will require investment and time, investing in your people to acquire the new skills, to address the change management, and to invest in and review your existing IT processes.
Unfortunately, while from the marketing perspective it sounds great, investment and time need to be put into place to support this trend and roll it out.
Gardner: Let me address for just a second. If we're in a recession and if you're no longer fighting the fires at the gross edge of the curve, which is to say, you are just trying to keep up with demand in your IT adoption patterns, but you've had a little bit of a lull in how much demand there is for products and services in your organization, wouldn't that be the perfect time to invest in your people and processes to elevate them to service orientation?
Bennett: Definitely. I actually see recession as an opportunity within IT, because it gives you opportunity to reset thinking and reset IT's approach to actually delivering IT to the business.
I completely agree that the availability of the cloud is a platform for doing this kind of thing. It's a very fortuitous coincidence with the business needs of enterprises that are short on capital.
Coffee: That's particularly true for the cloud, because you can do development without major capital investments. You can build systems, conduct pilot projects, and have solutions that are ready to go. They've been tested with the appropriate stakeholders, and when business conditions improve and create the opportunity to use those at larger scale, you can scale very rapidly to larger capacity, without needing to go through another round of capital budgeting.
So, yes, I completely agree that the availability of the cloud is a platform for doing this kind of thing. It's a very fortuitous coincidence with the business needs of enterprises that are short on capital.
Gardner: So, not only do you fix your roof when the sun is shining, but you can re-factor or modernize your roof while the sun is shining as well.
Coffee: Very much so.
Gardner: Chris Harding, what's going on in The Open Group's work group, when it comes to this skills and process level? Do you see that the recession and/or interest in cloud has increased the demand for people with architectural skills and perhaps certification that is a part and parcel with SOA?
Harding: It's difficult to measure the effect of the recession on education. You could say that if people aren't being required to do productive work 110 percent of the time, then they will give more thought on training and education.
But, also as far as SOA goes, we’ve gotten past the hype. We're into the steady take up deployment of SOA, and that's precisely the time when companies will be looking for people who can demonstrate their ability, professionalism, and qualification with this particular way of doing things.
Allocation of time
And, it's within the SOA Work Group. You can see that at our conferences. We spend a lot less time on presentations on how SOA is evolving and what the new things are, because really there isn't a lot that’s new coming up. We spend a lot more time putting together assessing tutorials on SOA practice and on the how the standards that the SOA Work Group is producing should be interpreted and used.
So, yes, we're moving very definitely into the period with SOA where establishing skills and certifying skills is going to be more important than developing new ways of doing things. We have those ways and we need to consolidate them and ensure that people can use those skills in their work and do so in a beneficial way.
With cloud, it's too early to do that yet, because we're still looking at the new developments when it comes to cloud.
Coffee: Dana, I’d agree with everything that Chris just said, except the statement about it being too early. When I talk with systems integration firms in India, like Tata and Infosys and Wipro, in Hong Kong with people like IBM, and in Europe with people like Accenture, all of them tell me that they are now actively engaging with customers on cloud-based options, as well as either on-premise based options or combinations of the two. So, I would say that we're well into the practitioner phase and not merely the conceptual phase.
It's very dangerous to oversell any of this as a proposition of cost reduction, when really it's an opportunity to reallocate resources from relatively low-value tasks, like administering a software stack, to higher-value tasks like reengineering a business process.
But, in every other respect, I completely agree with Chris. Chris and Steve have both talked about investments we'll be making in training and the development of new skills. This might make some people say, "Wait a minute, I thought this was about reducing my costs."
It's very dangerous to oversell any of this as a proposition of cost reduction, when really it's an opportunity to reallocate resources from relatively low-value tasks, like administering a software stack, to higher-value tasks like reengineering a business process.
If you want to have cheaper IT, you can have cheaper IT, but your company won’t be competitive with all of the performance and service that the industry has come to expect. Over time, IT budgets rise because the return on investment (ROI) is so compelling.
The combination of SOA, which makes your various business units able to cooperate more effectively, with cloud environments which allow you to handle very bursty workloads and conduct very cost-effective pilot projects and scale the ones that work very rapidly, increase the ROI of IT spending. Therefore, over time, IT budgets are going to continue to rise.
Bennett: From what I see, I agree with both of you. I see that there are a number of crossover skills, especially the non-technical skills. Both SOA and cloud heighten the need and importance of existing skills, such as governance, good design, portfolio management, and, more importantly, business architecture.
What people are talking about is the opportunity to redirect costs to area such as business architecture, and business architecture is part of enterprise architecture (EA). That's not purely IT focused, but the wider concern -- investing stuff like business capability maps to understand exactly where I should utilize SOA and cloud with my organization -- is going to be key.
Gardner: Now, on this economic issue, many of the executives I see aren't necessarily interested in reducing the IT budget dollar for dollar, but they are interested in reducing the percentage that IT consumes as a function of overall revenues and perhaps also profit. So if we are going to make things more productive, more efficient, even as we grow IT, what is it about SOA combined with cloud computing that could help do that?
Decentralization
Coffee: The sell with SOA very early on was that by making the IT environment more modular, it became more reasonable to have a business process owner be more intimately involved in the implementation of the technology that would execute that process.
You saw a decentralization, to some degree, of functions that had previously been monolithic development of massive applications, which was clearly the IT department’s charter, to more of a decentralized business-process construction task, which also included some development but now that became more exfiltrated out into the business units themselves.
So, the IT budget, as a line item, is not conspicuously bigger. In fact, it may actually shrink, because the IT department now is a composer and integrator of stuff that may now be getting done with the operating budget by personnel, who are on the payroll as members of a business unit, instead of members of an IT organization.
Gardner: As we also move culturally and organizationally, we'll probably see more adoption of shared services and management around the different economic models of how consumption is charged back. Therefore, with market forces as they tend to go, we’ll see higher efficiency. That is, you only pay for what you consume and you only consume that which you’re interested in paying for. If cloud computing allows for that elasticity, then you’ll only need to provision the resources required for those shared services.
This will, in turn, enable the consuming enterprises to concentrate on the things that they are particularly good at.
Now, this is a little bit of a vision, of course, but it strikes me that there is an opportunity for higher utilization, higher productivity, and therefore lower total cost. Any reaction?
Harding: That certainly must be one of the factors that will enable cloud computing to make enterprises more efficient -- the elasticity and the take-up effect. It also has a major effect on the risk that an enterprise needs to take on. But, there is a bigger factor, which is meant to drive down cost, and that is competition.
If you take service orientation and cloud in combination, you’re seeing the ability of people to buy services from different suppliers, for those suppliers to compete, and for those suppliers to concentrate on the services that they are particularly good at. This will, in turn, enable the consuming enterprises to concentrate on the things that they are particularly good at.
So, you don’t need to dissipate your efforts on running an inefficient IT department, which is not your core business. You can outsource that, get a specialist to do it much better, and concentrate on what you're good at. That is the real dynamic that will improve things economically.
Gardner: Do you think that a general adoption of services orientation will allow for those services to be procured based more on the service-level agreements (SLAs) and the requirements, regardless of who is actually sourcing them and building out the data centers? I think this makes a great deal of sense. How does that sit with you?
Harding: I don’t think that people will be switching from supplier to supplier to supplier on a per-transaction basis. They'll be establishing long-term business relationships and partnership with their service suppliers, and it will be in that context that they go out and get more services. But, the fact that they have a choice of suppliers to partner with will mean that those suppliers will be able to provide more efficiently, because they specialize in providing those services.
Now, from an Open Group perspective, there is a danger that you may become locked into a particular supplier. Part of our role in promoting open systems is to push for the standards to be in place so that that doesn’t happen. Provided we can prevent that locking, it’s altogether a very healthy situation.
Gardner: Peter Coffee, I’m sure you have some thoughts on this. It’s sort of the ecosystem question, and the ability to procure based on the differentiation of core and context services.
Granularity is surprising
Coffee: Certainly. The granularity of this marketplace is quite surprising to many people who haven’t looked at it closely. We see already people building applications, in which they have shopped the marketplace and found a cloud storage proposition from one provider, a cloud application development platform from another, social networking algorithms and facilities from yet a third provider and have built some really interesting strategic business solutions.
Starbucks is an example. They built applications that drew information from existing public service directories of community service opportunities, wrapped that in an application shell constructed with our Force.com facilities and packaged that in such a way that it’d be dropped into people’s Facebook pages to create community momentum around projects. They then built a site where, if you did five hours of community service, you could go to your local Starbucks and get a cup of coffee.
This was all implemented in matter of weeks to support President Obama’s call for renewed commitment to public service. It was very strategic from a marketing point of view, lovely from a technology point of view, and dirt-cheap, because there was no capital hardware involved. It was all a composition and integration of existing services from multiple providers. It’s quite startling to many people to realize what a supermarket of services has already come into being.
Gardner: This takes us to another level here, which would be not only thinking about how SOA and cloud come together to do things you had done in the past better, but perhaps be able to do things that you could not have been able to do at all before. We'll have to have this as our closing concept. I’d like to go around the panel. Chris Harding, what is it about SOA and cloud coming together that enables entirely new types of innovation and business benefit?
Harding: I don’t know that there is necessarily anything special, but maybe it’s because when you solve one set of problems you're able to turn your attention to other things and new things that pop up. It’s very difficult to foresee what new developments there will be, but you do get them, and it’s good news.
Those basic principles are going to allow us to take evolutionary technologies and approaches and probably revolutionize the way that IT actually interacts with the business.
Gardner: At least we move from "if you build it," the productivity, services, or innovation will come to "let’s just go and do the innovation" and we’ll find a way to do it without having to build very much or anything at all.
Harding: Right.
Gardner: Stephen Bennett, how about you? Do you see an opportunity for SOA and cloud to come together to enable entirely new ways of doing business?
Bennett: I won’t say new ways of doing business, but ways that we were trying to achieve probably over the last 5-10 years. The combination of cloud and SOA obviously brings together kind of speed and modularity. Those basic principles are going to allow us to take evolutionary technologies and approaches and probably revolutionize the way that IT actually interacts with the business.
So, in terms of IT being siloed -- "please develop and look after this application" -- it’s going to be more a move towards collaboration of how we can actually deliver business solutions to the ever-changing business dynamics.
Gardner: That really integrates IT more deeply into the business, I suppose, right?
Bennett: Exactly. It's being invited to the table at long last.
Gardner: And, Peter Coffee, the last word to you. What is your perception on being able to do things that hadn’t been done or been able to be done before with this unique combination and interplay between SOA and cloud in its various forms?
Coffee: Steve is exactly right that many of the concepts that are now receiving tremendous attention are not new concepts. I distinctly remember discussions almost 30 years ago about modular applications composed of various components, developed by different parties with different expertise, coming together for the first time in a connected environment. This was going to be one of the things that object-oriented modules of development were going to bring to us.
Very resilient
Finally, we have an environment in which connectivity and real-time linkage and integration of data and function instead of being costly, brittle, and time-consuming are now nearly free, very resilient, and can be done almost more quickly than they can be described.
This means that people are going to be doing more challenging work and working more closely with business units instead of having their time consumed by arduous, necessary, but relatively low-value tests of infrastructure maintenance.
So the ROI will rise. The relevance to the business of IT will increase. The sophistication of the skills of the person who does IT for a living will be greater 10 years from now than it was 10 years ago or even today, but we’ll all be pretty happy with the results.
This means that people are going to be doing more challenging work and working more closely with business units.
Gardner: It strikes me that perhaps the perception of IT itself might benefit in that IT could be perceived as the enabler of innovation that others will actually do rather than IT being perceived as a bottleneck of innovation that IT needs to have enabled themselves.
Coffee: I've also heard it suggested that we should evolve the title of Chief Information Officer, which refers to a relatively static asset, to the title of Chief Process Officer, which suggests someone who is much closer to the present and future of the business.
Gardner: Or we could go Chief Innovation Officer and we won’t have to change the stationery.
Well, thanks so much. We’ve been listening here to a panel of experts talking about the interplay between SOA and cloud computing. I want to thank our panelists Dr. Chris Harding, Director of the SOA Work Group at The Open Group. Thank you, Chris.
Harding: Thank you, Dana. It’s been a pleasure.
Gardner: Steven G. Bennett, Senior Enterprise Architect at Oracle. Thank you, Stephen.
Steven G. Bennett: Thank you very much.
Gardner: Peter Coffee, Director of Platform Search for salesforce.com. Thanks so much, Peter.
Coffee: Enjoyed the conversation, thank you.
Gardner: You’ve been listening to a sponsored BriefingsDirect podcast that’s been done in conjunction with The Open Group’s Enterprise Architecture Practitioners Conference held in Seattle the week of February 1, 2010.
This is Dana Gardner, principal analyst at Interarbor Solutions. Thanks for listening and come back next time.
Listen to the podcast. Find it on iTunes/iPod and Podcast.com. Download the transcript. Sponsor: The Open Group.
Transcript of a BriefingsDirect sponsored podcast on the intersection of SOA and cloud computing recorded live at The Open Group’s Enterprise Architecture Practitioners Conference in Seattle. Copyright Interarbor Solutions, LLC, 2005-2010. All rights reserved.
You may also be interested in:
