SOA has been around for years but why is there a sudden growth in popularity and why should businesses implement a SOA architecture? Or should we say, what are the implications of a business not adopting SOA?
SDA Asia sits down with Ross Altman, Cheif Technology Officer of SOA and Business Integration in Sun Microsystems to get the answers.
SDA: How do you see the SOA landscape in Asia today?
Ross Altman (RA): To understand the SOA landscape in Asia one must first decide where SOA would most likely be attractive to folks. That’s an interesting question because SOA is clearly a technology initiative rather than a business specific initiative. It allows us to have capabilities to build and apply applications in a more effective way and to build a certain type of application. I call these composite applications.
If you consider the challenges of building a composite application, this is an application that represents some new logic but we are leveraging on some existing systems in order to deliver that new logic to you. This means we have to integrate existing applications that weren’t meant to be integrated in order to create composite applications.
The real value of SOA is in composite application. When people see the value of building a composite application they will see the value in using SOA. Otherwise, it’s just an exercise in academic experimentation-- it’s the using of a new technology in order to experiment with it and doesn’t really deliver any value to the organisations.
SDA: Although SOA has entered the Asian region, do you think it has made its mark in the region?
RA: SOA is making its mark to varying degrees. Some companies have been very aggressive to use SOA and for that reason they are better suppliers and better consumers on this B2B SOA, composite application scenario. That’s where we are going to see the people who have the best benefit from SOA.
The industries that are doing the most with it are manufacturing. You see a lot of this in banking because in banking we’ve gone into straight through processing and beyond straight through processing. Insurance companies do it for claims adjustment. It’s used in telcos. A lot of the really cool and neat telco services that you get on your device is SOA. They can build services from the network such as location services, services from the other applications e.g a database of restaurant reviews/locations, and it combines all of that to offer a new service to a customer. This is one of the ways that mobile providers are trying to increase account retention and be more successful in their markets.
So you can see that in communication, manufacturing, financial services, healthcare etc. Almost any area that you might want to look at there is going to be some requirement for consolidating data. I’d be hard pressed to think of any industry where there wasn’t a really good reason to use SOA.
SDA: In Asia, which regions have you noticed as being early adopters of SOA and which regions are trailing behind? Why do you think this is the trend?
RA: The uptake of this technology in Asia varies, from country to country, from market to market, from organisation to organisation, mostly based on their willingness to adopt new technology and their aggressiveness in adopting new business models because again it’s the business practices, the business models, the business goals that drive IT investment.
I see a lot of uptake in China and India, no surprise because of outsourcing in both the manufacturing kind and business process outsourcing kind.
It’s early in the adoption cycle but I see a lot of interest in Thailand, Malaysia, Singapore and especially in Korea because it does a lot of manufacturing and in Japan, because of financial services.
The biggest indicator of which countries are going to be more or less aggressive is to possibly look at the baseline level of industrialization and also look at the relative degree of economic growth, companies is those countries are investing heavily.
SDA: SOA has been a buzzword for the longest time now, but whilst there are many organisations who have heard of SOA, most of them wouldn’t even know where to begin to implement one. What advice would you give an organisation interested in embarking on an SOA project?
RA: SOA has to be implemented in a pragmatic way. You can take an academic approach but obviously there are a lot of drawbacks that go in to gold plate your plans. There is a certain amount of top down planning that’s involved. You have to have an architecture where everybody shares the same concept, as they say, share the same vocabulary. So that when we talk about what we’re doing we can use the same terms and we know what it means.
It’s useful to have some technology involved although this is not primarily a technology problem. It’s more of an organisational issue that is supported by good software. So you need an architecture, you need to plan, you need a methodology. If you’ve got a methodology for doing applications development today that’s good. You probably will have to make some adjustments to it to support the SOA composite applications development. So an adjusted, updated methodology is important.
They also have to take into consideration certain governance models. One of which is “who pays for it?” Somebody has to sit down and decide how we are going to pay for this and it does not just mean, where are we going to find the money but how do we actually allocate funds and account for the relationship between investment and the lines of business that leverage that investment. Other governance models include design time governance and run time governance which is service policy enforcement.
So this is what goes into the making of a good SOA. There is a lot of planning. If you think it’s simple than you are going to be really disappointed because there is nothing anybody can do to make this simple. This is actually a little bit more complex than building discreet, standalone, boxed applications but, its cheaper and it gives you more flexibility in the long run and you won’t be able to compete in the long run if you build boxed, standalone applications.
SOA: According to Gartner, whilst many companies recognize the benefits of SOA, the lack of governance is one of the most common reasons why many implementations fail. Do you agree with this statement?
RA: Governance is something organisations don’t see the need for immediately.
They don’t realise that they can’t just go ripping into it and start implementing SOA. In which case you will get some services built, and you will get a little bit of reuse. But because you didn’t plan and you didn’t implement the model for design time and run time governance you’re really not getting all the reuse, all the manageability, all the control that you’re going to need.
You will be more competitive than the companies that stuck to their old monolithic approach, but you won’t be as competitive as the guys who implemented a more flexible approach to SOA. It’s entirely possible to implement SOA poorly and if you do, you will probably be disappointed with the results. There are no absolutes anywhere and there are no absolutes here. But it is possible to do a better job or a less impressive job in implementing SOA.
SDA: So how do you think we can do a better job with Governance?
RA: Hire experienced professionals to get you started so that you make the necessary upfront investment in architecture, methodology and governance models. Then start a bottom up development approach. I think that gives you the best opportunity to be successful. Don’t spend too much time planning because then people will get impatient and your upfront cost will start increasing and it will get increasingly harder to work off those upfront cost.
SDA: Could you tell us more about Sun’s “4 Ps” strategic approach to SOA?
RA: The 4 Ps are practice, process, program, people and cover all aspects of the organisational development to ensure organisations have the right plan in place.
Organisations have to plan and keep a process in place. Then they have to train the people properly so they have the skill to execute the process and implement the plan.
SDA: SOA has been advocated by major software players like Oracle and IBM and Sun is considered as quite a late entrant into this market. Can you comment on this?
RA: Sun has been talking about SOA but we didn’t have a product. In the market place people tend to not pay too much attention to best practices and the “4Ps” but they do spend a lot of time looking at products. SOA is the “4Ps”. Product isn’t one of them and that should be the last one should worry about.
So to get a product in place, Sun decided to acquire SeeBeyond whose offerings dovetail perfectly with Sun’s and we now collectively have an impressive corporate resume in delivering SOA in the most effective way possible.
SDA: How are you planning to take on your competitors who have already established themselves in this space? What would you say is your competitive edge?
RA: We compete with them everyday.
Sun started in 2001 to design a comprehensive environment that will support composite applications on top of service oriented architecture. We called the product ICAN--integrated, composite application, network. We recognised that the objective of the investment is to support composite applications, not just to build a bunch of services.
Having a comprehensive tool set is very important. A comprehensive tool set involves a lot of things. It involves EAI technology, BPM technology, B2B technology etc. You need to have all that stuff in order to be able to support services and composite applications.
When we came out with ICAN in Oct 2003, at that point we were the only guys out there with a comprehensive tool set. All our competitors decided to catch up by making a bunch of acquisitions. But onlhy ended up with a bunch of tools built at different times in different architectures in different compositories.
Sun, on the other hand, has one tool set, built at one time by one group of developers with one architecture. Being comprehensive is no longer differentiating. Being comprehensive and fully integrated is.
One of the things that separate us from a lot of our competitors is our toolsets built to implement an open standard for communications of our components and we build out that toolset using an open source development model. We have an open source community. There are hundreds of contributors to that community and there are dozens of components that have been built and made available through that community.
Our leverage on this open source community means that we can compete really well with the guys that don’t use open source.
SDA: How promising a future do you think SOA has and where do you see SOA 5 years from now?
RA: I think the companies that don’t do SOA won’t be successful and the companies that do, will. The notion of a vertically integrated company was over some 100 years ago.
You can’t build a vertically integrated company, it has to be flexible. How do you do that? SOA!
Everyone will be doing SOA going forward.
SOA will be more or less the same 5 years from now. It won’t be radically different. But we will get better at implementing it from a practice, process, program, people point of view. Our tools will get better. There are other related parts that we want to get good at also like event processing. |
