Edited transcript of weekly BriefingsDirect[TM] SOA Insights Edition podcast, recorded
Listen to the podcast here.
Dana Gardner: Hello, and welcome to the latest BriefingsDirect SOA Insights Edition, Volume 27. A weekly discussion and dissection of Services Oriented Architecture (SOA) related news and events with a panel of industry analysts, experts and guests.
I'm your host and moderator, Dana Gardner, principal analyst at Interarbor Solutions. We’re joined today by a handful of prominent IT analysts who cover SOA and related areas of technology, business, and productivity.
Topics we're going to discuss this week include the SOA & Business Process Conference held by Microsoft in
We're also going to discuss issues around "Green SOA." How will SOA impact companies, as they attempt to decrease their energy footprint, perhaps become kinder and gentler to the environment and planet earth, and what SOA might bring to the table in terms of a long-term return on investment (ROI), when energy related issues are factored in?
Jim Kobielus: Hi, Dana. Hello, everybody.
Neil Macehiter: Hi, Dana. Hi, everyone.
Joe McKendrick: Thanks, Dana, glad to be here.
On Microsoft-Oriented Architecture and the SOA Confab ...
It's going to be several years out before these products arrive -- we probably won’t even see data until well into 2008 on a number of these products. Part of the logic seems to be that you can write anywhere, have flexibility in your tooling, and then coalesce around a variety of models or modeling approaches to execute through an über or federated modeling approach that Microsoft seems to be developing. That would then execute or deploy services on Microsoft foundational infrastructure.
I'm going to assume that there is also going to be loosely coupled interoperability with services from a variety of different origins and underlying infrastructure environments, but Microsoft seems to be looking strategically at this modeling layer, as to where it wants to bring value even if it’s late to the game.
Let’s start with Jim Kobielus. Tell us a little bit about whether you view Microsoft's moves as expanding on your understanding of their take on SOA, and what do you make of this emphasis on modeling?
Kobielus: First, the SOA universe is heading toward a model-driven paradigm for distributed service development in orchestration, and that’s been clear for a several years now. What Microsoft has discussed this week at its SOA and BPM conference was nothing radically new for the industry or for Microsoft.
Over time, with Visual Studio and the .NET environment, they've been increasingly moving toward a more purely visual paradigm. "Visual" is in the very name of their development tool. Looking at the news this week from Microsoft on the so-called Oslo initiative, they are going to be enhancing a variety of their Visual Studio, BizTalk Server, BizTalk Services, and Microsoft System Center, bringing together the various metadata repositories underlying those products to enable a greater model-driven approach to distributed development.
Kobielus: Yes, BizTalk Server for premises-based and BizTalk Services, software as a service (SaaS), the channel through which it can deliver BizTalk functionality going forward. I had to pinch myself and ask myself what year this is. Oh, it’s 2007, and Microsoft is finally getting modeling religion. I still remember in 2003-2004 there was a big up swell of industry interest in model-driven architecture (MDA).
Kobielus: I was thinking, okay, that’s great, Microsoft, I have no problem with your model-driven approach. You're two, three, or four years behind the curve in terms of getting religion. That’s okay. It’s still taking a while for the industry to completely mobilize around this.
In order words, rather than developing applications, they develop business models and technology models to varying degrees of depth and then use those models to automatically generate the appropriate code and build the appropriate sources. That’s a given. One thing that confuses me, puzzles me, or maybe just dismays me about Microsoft’s announcement is that there isn't any footprint here for the actual standards that have been developed like OMG’s unified modeling language (UML), for example.
Microsoft, for some reason I still haven’t been able to divine, is also steering clear of UML in terms of their repositories. I'm not getting any sense that there is a UDDI story here or any other standards angle to these converged repositories that they will be rolling out within their various tools. So, it really is a Microsoft Oriented Architecture. They're building proprietary interfaces. I thought they were pretty much behind open standards. Now, unless it’s actually 2003, I have to go and check my calendar.
Some of the infrastructure, be it repository, ESB, management, or governance, needs to be quite open. So, you're saying you're not sure that you're seeing that level of openness. It reminds us of the CORBA versus COM and DCOM situation. OMG was involved with that and supported the development of CORBA
Let’s go to Neil Macehiter. Do you see this as MOA or do you think that they are going to have to be open, if it’s going to be SOA values?
Macehiter: I don’t see this as exclusively Microsoft-oriented, by any stretch. I’d also question Jim’s comment on there being nothing radically new here. There are a couple of elements to the strategy that Microsoft’s outlined that differentiate it from the model-driven approaches of the past.
The first is that they are actually encompassing management into this modeling framework, and they're planning to support some standards around things like the service modeling language (SML), which will allow the transition from development through to operations. So, this is actually about the model driven life cycle.
The second element where I see some difference is that Microsoft is trying to extend this common model across software that resides on premises and software that resides in the cloud somewhere with services. So, it has a common framework for delivering, as Microsoft refers to it, software plus services. In terms of the standard support with respect to UML, Microsoft has always been lukewarm about UML.
A few years ago, they were talking about using domain specific language (DSL), which underpin elements of Visual Studio that currently exist, as a way of supporting different modeling paradigms. What we will see is the resurgence of DSL as a means of enabling different modeling approaches to be applied here. The comment regarding UDDI is only one element at the repository, because where Microsoft is really trying to drive this is around a repository for models, for an SML model or for the models developed in Visual Studio, which is certainly broader.
Macehiter: No, so this smacks of being a very ambitious strategy from Microsoft, which is trying to pull together threads from different elements of the overall IT environment. You've got elements of infrastructure as a service, with things like the BizTalk Services, which has been the domain of large Web platforms. You've got this notion of computer applications in BPM which is something people like IBM, BEA, Software AG, etc. have been promoting.
Microsoft has got a broad vision. We also mustn’t forget that what underpins this is the vision to have this execution framework for models. The models will actually be executed within the .NET framework in the future iteration. That will be based on the Window’s Communication Foundation, which itself sits on top of the WS-* standards and then also on top of Windows Workplace Foundation.
So, that ambitious vision is still some way off, as you mentioned -- beta in 2008, production in 2009. Microsoft is going to have to bring its ISVs and systems integrator (SI) community along to really turn this from being an architecture that's oriented towards Microsoft to something broader.
Gardner: Now, Neil, if Microsoft is, in a sense, leapfrogging the market, trying to project what things are going to be several years out, recognizing that there is going to be a variety of modeling approaches, and that modeling is going to be essential for making SOA inclusive, then they are also going to be federating, but doing that vis-à-vis their frameworks and foundations.
If there is anything in the past that has spurred on industry standards, it's been when Microsoft puts a stake in the ground and says, “We want to be the 'blank,'” which, in this case, would be the place where you would federate models.
Kobielus: I’m glad you mentioned the word "federation" in this context, because I wanted to make a point. I agree with Neil. I’m not totally down on what Microsoft is doing. Clearly, they had to go beyond UML in terms of a modeling language, as you said, because UML doesn’t have the constructs to do deployment and management of distributed services and so forth. I understand that. What disturbs me right now about what Microsoft is doing is that if you look at the last few years, Microsoft has gotten a lot better when they are ahead of standards.
When they're innovating in advance of any standards, they have done a better job of catalyzing a community of partners to build public specs. For example, when Microsoft went ahead of SAML and the Liberty Alliance Federated Identity Standards a few years back, they wanted to do things that weren't being addressed by those groups.
Microsoft put together an alliance around a spec called WS-Federation, which just had sort of hit-and-miss adoption in the market, but there has been a variety of other WS-* standards or specifications that Microsoft has also helped to catalyze the industry around in advance of any formal, de-jure standard. I'd like to see it do the same thing now in the realm of modeling.
Macehiter: My guess is that’s exactly what they’re doing by putting a stake in the ground this early. "This is coming from us. There are going to be a lot of developers out there using our tools that are going to be populating our repositories. If you're sensible, you're going to federate with us and, therefore, let’s get the dialogue going." I think that’s partly why the stake is out there as early as it is.
Gardner: Let’s go to Joe McKendrick, Joe, we've seen instances in the past where – whether they're trailing or leading a particular trend or technology -- Microsoft has such clout and influence in the market that they either can establish de-facto standards or they will spur others to get chummy with one another to diminish the Microsoft threat. Do you expect that Microsoft's saying they're going to get in the modeling federation and repository business will prompt more cooperation and perhaps a faster federated standard approach in the rest of the market?
McKendrick: Definitely more and more competitive responses. Perhaps you’ll see IBM, BEA, Oracle, or whatever other entity propose their own approaches. It's great that Microsoft is talking SOA now. It's only been about a year that they have really been active.
McKendrick: I think what's behind this is that Microsoft has always followed the mass market. Microsoft’s sweet spot is the small- and the medium-business sector. They have a presence in the Fortune 500, but where they’ve been strong is the small to medium businesses, and these are the companies that don’t have the resources to form committees and spend months anguishing over an enterprise architectural approach, planning things out. They may be driven by the development department, but these folks have problems that they need to address immediately. They need a focus and to put some solutions in place to resolve issues with transactions, and so forth.
However, there are still very few Fortune 500 or Global 2000 companies that are pure Microsoft shops. In many respects, enterprise Java, distributed computing, and open-standards approaches have dominated the core environment in architecture for these larger enterprises. If Microsoft is going to get into SOA, they're in a better position to do what we’ve been calling Guerrilla SOA, which is on a project-by-project basis.
If you had a lot of grassroots, small-developer, department-level server-oriented activities that Microsoft infrastructure would perhaps be better positioned to be dominant in, then that’s going to leave them with these islands of services. A federated modeling level or abstraction layer would be very fortuitous for them. Anyone have any thoughts about the comprehensive enterprise-wide SOA approach that we have heard from others vendors, versus what Microsoft might be doing, which might not be comprehensive, but could be in a sense grassroots even within these larger enterprise.
Macehiter: The other vendors in the non-Microsoft world might talk about enterprise-wide SOA initiatives and organizations that are planning to adopt SOA on an enterprise-wide basis, based on their infrastructure. The reality is that the number of organizations that have actually gone that far is still comparatively small, as we continually see with the same case-study customers being reintroduced again and again.
Microsoft will have to adopt an alternative model. For example, I think Microsoft will follow a similar model and explore the base they had around the developer community within organizations with things like Visual Studio.
SQL Server is pretty well deployed in enterprise elements of the application platform, by virtue of their being bundled into the OS already. So, they're quite well-positioned to address these departmental opportunities, and then scale out.
This is where some of the capabilities that we talked about, particularly in combination with things like BizTalk Services, allow organizations to utilize workflow capabilities and identity management capabilities in the cloud to reduce the management overhead. The other potential route for Microsoft is through the ISV community.
Gardner: I suppose one counterpoint to that is that Microsoft is well positioned with it's tools, frameworks, skill set, and entrenched positions to be well exploited for creating services, but when it comes to modeling business processes, we're not really talking about a Visual Studio-level user or developer. Even if the tools are visually oriented, the people or teams that are going to be in a position to construct, amend, develop, and refine these business processes are going to be at a much higher level. They're going to be architects and business analysts. They're going to be different types of persons.
They are going to be people who have a horizontal view of an entire business process across the heterogeneous environments and across organizational boundaries. Microsoft is well positioned within these grassroots elements. I wonder if they can, through a modeling federation layer and benefit, get themselves to the place where they are going to be the tools and repository for these analysts and architect level thinkers.
Kobielus: I think they will, but they need to play all this
Macehiter: The other community there would be the SIs, who do a lot of this work on behalf of organizations. As part of the
McKendrick: As an addendum to what you just said, Microsoft isn't necessarily going to go directly after customers of IBM, BEA, etc. IBM is providing this potential to companies that have been under-served, companies that cannot afford extensive SOA consulting or integration work. It's going after the SMB sector, the
McKendrick: Exactly, Microsoft is the disrupter in this case. There are other markets where Microsoft is being disrupted by Web 2.0, but in SOA, Microsoft is playing the role of disrupter and I think that’s what their strategy is.
Kobielus: I want to add one last twist here. I agree with everything Joe said. Also, the
What Microsoft really needs to be feature-competitive in the MDM market is a model-driven, visual business-process development and stewardship tool. That way teams of business and technical analysts can work together in a customer data-integration, product information-management, or financial consolidation hub environment to build the complex business logic into complex applications under the heading of MDM. If Microsoft's MDM team knows what they are doing and I assume they do, then they should definitely align with the Oslo initiative, because it will be a critical for Microsoft to compete with IBM and Oracle in this phase.
We've heard a little bit about a forthcoming protocol -- BizTalk Services 1 -- and that probably will relate to Microsoft's Live and other online-based approaches. The end user, be they an architect and analyst or someone who is going to be crafting business processes, if they're using a strictly Web-based approach, they don’t know or care what’s going on beneath the covers in terms of foundations, frameworks, operating systems, and runtime environments. They are simply looking for ease of acquisition and use productivity-scale reliability.
It strikes me that Microsoft is now working towards what might be more of a Salesforce.com, Google, or Amazon-type environment, where increasingly SOA is off the wire entirely. It really is a matter of how you model and tool those services that becomes a king maker in the market. Any thoughts on how Microsoft is positioning these products for that kind of a play?
Macehiter: Definitely. The software plus services, which is the way that Microsoft articulates this partitioning of capability between on-premise software and services delivered in the cloud, is definitely a key aspect of the
For example, if an organization needs to do some message flow that crosses between organizations over the firewall, BizTalk Services will provide a capability that allows you to explore that declaratively. You can see that evolving, but that’s more infrastructure services. Clearly, another approach might be a high-level service, an application type service, and this architecture that Microsoft is talking about is attempting to address that as well.
This is definitely a key element of the story, which is about making sure that Microsoft remains relevant in the face of in an increasing shift, particularly in the SMB market, towards services delivered in the cloud. It’s about combining the client, the server and services, and providing models in terms the way you think about the applications you need and in terms of the way you manage and deploy them that can encompass that in a way that doesn’t incur significant effort.
Macehiter: Yes. That’s why this is a better attempt by Microsoft to change the game and push the boundaries. It’s not just a model-driven development revisited in a NDI and a .NET world. This is broader than that.
McKendrick: Forming the mass market, exactly.
McKendrick: And reaching out to companies that could not afford SOA or comprehensive SOA, which it's done in the past.
On Green SOA and the IT Energy-Use Factor ...
It's a very big topic today. In fact, it was the top of the strategic technology areas that Gartner Group identified for 2008. Green IT was named number one, a top-ten strategic technology area. How does SOA impact this? Jim Kobielus, you have been given this a lot of thought. Give us the lay of the land.
Kobielus: Thank you, Dana. Clearly, in our culture the Green theme keeps growing larger in all of our lives, and I'm not going to belabor all the ramifications of Green. In terms of Green, as it relates to SOA, you mentioned just a moment ago, Dana, the whole notion of SOA is based on abstraction, service contracts, and decoupling of the external calling interfaces from the internal implementations of various services. Green smashes through that entire paradigm, because Green is about as concrete as you get.
SOA focuses on maximizing the sharing, reuse, and interoperability of distributed services or resources, application logic, or data across distributed fabrics. When they're designing SOA applications, developers aren't necessarily incentivized, or even have the inclination, to think in terms of the ramifications at the physical layer of these services they're designing and deploying, but Green is all about the physical layer.
In other words, Green is all about how do human beings, as a species, make wise use and stewardship of the earth’s nonrenewable, irreplaceable resources, energy or energy supplies, fossil fuels, and so forth. But also it’s larger than that, obviously. How do we maintain a sustainable culture and existence on this planet in terms of wise use of the other material resources like minerals and the soil etc.?
Kobielus: Yes, first and foremost, it’s pitched at the energy level. In fact, just this morning in my inbox I got this from IBM: "Join us for the IBM Energy Efficiency Certificate Announcement Teleconference." They're going to talk about energy efficiency in the datacenter and best practices for energy efficiency. That’s obviously very much at the core of the Green theme.
Now, getting to the point of how SOA can contribute to the greening of the world. SOA is the whole notion of consolidation -- consolidation of application logic, consolidation of servers, and consolidation of datacenters. In other words, it essentially reduces the physical footprint of the services and applications that we deploy out to the mesh or the fabric.
Kobielus: Well, SOA is predicated on sharing and reuse. Okay, your center has a competency. You have one hunk of application logic that handles order processing in the organization. You standardize on that, and then everybody calls that, invokes that over the network. Over time, if SOA is successful other centers of development or other deployed instances of code that do similar things will be decommissioned to enable maximum reuse of the best-of-breed order-processing technology that’s out there.
As enterprises realize the ROI, the reuse and sharing should naturally lead to greater consolidation at all levels, including in the datacenter. Basically, reducing the footprint of SOA on the physical environment is what consolidation is all about.
Gardner: So, these trends that are going concurrently -- unification, consolidation, and virtualization -- allow you to better exploit those activities and perhaps double down on them in terms of a fewer instances of an application stack, but more opportunity to reuse the logic and the resources more generally. So a highly efficient approach that ultimately will save trees and put less CO2 in the atmosphere.
Kobielus: I want to go back to Microsoft. Four years ago, in 2003, I went to their analyst summit in
That’s kind of interesting. They are making a whack at both trying to model from the application down to the physical layer and think through the physical consequences of application development activities.
That will ultimately help the planet, as well, because we wouldn’t have umpteen datacenters in every single company of more than a 150 people. We could start centralizing this almost like a utility would. We would think that these large companies, as they put in these massive datacenters, could have the opportunity for a volume benefit in how they consume and purchase energy.
Macehiter: We need to step back and look at what we are talking about. You mentioned ROI. If we look at this from a Green ROI perspective, organizations are not going to be looking at SOA as the first step in reducing their Green footprint. It's going to be about server and storage consolidation to reduce the power consumption, provide more efficient cooling, and management approaches to ensure that servers aren’t running when they don’t need to be. That’s going to give them much bigger Green bang for the buck.
Certainly, the ability to reuse and share services is going to have an impact in terms of reducing duplications, but in the broader scheme of things I see that contribution as being comparatively small. The history that we have is largely ignoring the implications of power and heat, until we get to the size of a Google or a Microsoft, where we have to start thinking about putting our datacenters next to large amounts of water, where we can get hydroelectric power.
So, IT has a contribution to make, but there isn't anything explicit in SOA approaches, beyond things like service reuse and sharing that can really contribute. The economies of scale that you get from SaaS in terms of exploiting those services come from more effective use of the datacenter resources. This is those organizations' business, and, given the constraints they operate under, they can’t get datacenters big enough, because then there are no power stations big enough.
McKendrick: In an indirect way, it sounds right, but I want to take an even a further step back and look at what we have here. Frankly, the Green IT initiative is misguided and the wrong questions are being asked about Green IT. Let me say that I have been in active environmental causes and I have done consulting work with a company that has worked with utilities and ERP Electric Car Research Institute on energy saving initiatives.
It's great that IT is emphasizing efficient datacenters, but what we need to look at is how much energy IT has saved the world in general? How much power is being saved as a result of IT initiatives? SOA rolls right into this. For example, how many business trips are not being taken now, because of the availability of video conferencing and remote telecommuting, telework and things of that sort? We need studies. I don’t have the data on this and there isn’t any data out there that has really tracked this. In e-commerce, for example, how many stores have not been built because of e-commerce?
Macehiter: The reality is that we can’t offset what we’ve saved in the past against what we are going to conceive in the future. We are at a baseline and it is not about apportioning blame between industries and saying, "Well, IT doesn’t have to do so much, because we’ve done a lot in the past."
McKendrick: But, we are putting demands on IT, Neil. We're putting a lot of demands on IT for additional IT resources.
Macehiter: If you go into a large investment bank, and look at what proportion of their electricity consumption is consumed by IT, I'd hazard a guess that it's a pretty large chunk, alongside facilities.
McKendrick: And probably lots of demands are put on those datacenters, but how much energy is that saving because of additional services being put out to the world, being put out to society?
McKendrick: The problem is, Dana, we don’t know. There are no studies. I'd love to see studies commissioned. I'd love to see our government or a private foundation fund some studies to find out how much energy IT has been saving us.
Kobielus: I agree with everything you guys are saying, because the issue is not so much reducing IT’s footprint on the environment. It’s reducing our species' overall footprint on the resources. One thing to consider is whether we have more energy-efficient datacenters. Another thing to consider is that, as more functionality gets pushed out to the periphery in terms of PCs and departmental servers, the vast majority of the IT is completely outside the datacenter.
Macehiter: Let's not forget that this isn't just about enterprise use of IT. If I look at my desk, as a consumer of IT, I've got a scanner, hard disk, two machines, screen, two wireless routers, and speakers that are all consuming electricity. Ten years ago, I just wouldn’t have had that. So, we have to look broader than the enterprise. We can get into a whole other rat’s nest, if we start into safe nuclear power or having wind farms near our datacenter.
Kobielus: In the Wall Street Journal this morning, there was an article about Daylight Saving Time. This year, in the
The article said that nobody has really done a study to show whether we are actually saving any energy from Daylight Saving Time? There have been no reliable studies done. So, when the legislatures change these weeks, they're just assuming that, by having more hours of daylight in the evening, we are using less illumination, therefore the net budget or net consumption of energy goes down.
In fact, people have darker mornings, and people tend to have more morning-oriented lives. People in the morning quite often are surfing the Web, and viewing the stuff on their TiVo, etc. So, net net, nobody even knows with Daylight Saving Time whether it's Green friendly, as a concept.
McKendrick: The efficiencies gained by IT might be outstripping any concerns about green footprints with datacenters. We need data. We need studies to look at this side of it. The U.S. Congress is talking about studying the energy efficiency of datacenters, and you can imagine some kind of regulations will flow from that.
Kobielus: I'm going to be a cynic and am just going to guess that large, Global 2000 corporations are going to be motivated more by economics than altruism when it comes to the environment. So back to the announcement today, on Nov. 2, about IBM launching an initiative to give corporate customers a way to measure and potentially monetize energy efficient measures in their datacenters.
I think IBM is trying to come up with the currency of sorts, a way to earn energy-efficient certificates that can then apply some kind of an economic incentive and/or metric to this issue. As we discussed earlier, the Green approach to IT might actually augment SOA, because I don’t think SOA leads to Green, but many of the things you do for Green will help people recognize higher value from SOA types of activities.
Kobielus: Thank you, Dana. I enjoyed it as always.
Macehiter: Thanks, Dana. Thanks, everyone.
McKendrick: Thanks, Dana. It was great to be here.
Produced as a courtesy of Interarbor Solutions: analysis, consulting and rich new-media content production.
If any of our listeners are interested in learning more about BriefingsDirect B2B informational podcasts or to become a sponsor of this or other B2B podcasts, please fill free to contact Interarbor Solutions at 603-528-2435.
Transcript of BriefingsDirect SOA Insights Edition podcast, Vol. 27, on Microsoft SOA and Green IT. Copyright Interarbor Solutions, LLC, 2005-2007. All rights reserved.