What Improvements Would You Like to See With SOA?

SOA in IT, as an integrated technology, is catered for well enough.

Improvements for Enterprise SOA do not come from SOA itself but from disciplines like business, information and Enterprise Architecture which are required to define the components that may become services, the information that is exchanged and the resources that implement the services. In their absence, SOA has to these architecture job before even starting to deal with services.
That is why SOA, at the Enterprise level, still fails.

Well, we read a lot about the convergence of SOA and Cloud Computing. There's an excellent book on the topic by David Linthicum and even I have written an article on the topic right here on ebizQ.

So, the one improvement I yearn for is the cementing and formalization of the relation between Cloud Computing and SOA. Today, cloud computing has three main delivery models: Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS), and Software-as-a-Service (SaaS). I propose including a fourth model called Information-as-a-Service.

Information-as-a-Service is the delivery model that formalizes the relation between SOA and Cloud Computing. It is a delivery model that is enabled by SOA at its core just as Virtualization is a key underpinning of IaaS. In other words, SOA is to Information-as-a-Service what Virtualization is to Infrastructure-as-a-Service. Put yet another way, Cloud Computing is the overall enterprise architecture pattern; SOA is the architectural pattern that enables one model of the overall Cloud: Information-as-a-Service.

Here is my wish list-

1. Greater focus on results, less on theory
2. Proactive 'Portfolio' approach to defining services, rather than collecting them piecemeal into a 'Library'
3. Service lifecycles based on governance priorities

Architecturally, we need better understanding of and agreement on how EDA and SOA fit together.

In order to look for improvemnts we should understand what went wrong in so many unsuccessful SOA initiatives.
I do agree with Jingesj Shah that SOA without EDA is one of the barriers and I also agree that non-pragmatic approach (theory instead of foucus on results) are major reasons for SOA initiatives failures and therefore SOA improvements should address these topics.
A major drawback of current SOA is the lack of coherent set of SOA Aanalysis and Development tools, including Composition and De-Composition capabilities.A mix of UML tools, Contextual Business Analysis tools, Capabilities Analysis and Service usage workshops is not good enough.

My one wish is that we could have definitive standards associated with SOA. As it stands now, SOA in terms of implementing services doesn't give us a huge amount as any well programmed system will have been developed using proprietary services for many years. On the other hand services implemented using standards (my own favourite being SOAP) enhance the propspects of services being reused exponentially as they can be reused from multiple languages or technologies.

I think the next 'frontier' for SOA is to address the business space, in particular how we model, realise and operate business services that are present as both self executed and partner capabilities. As Tarak says, SOA is now indivisible from the cloud - neither can reach its potential without the other - and so major areas of research need to cover business capability modelling (along with service levels, contract, pricing and composition) and the cloud platforms required to support the realistion of such capabilities and the formation of trusted cloud value webs that span many organisations.

I put a bit more context here: http://wp.me/p3Ou1-41 but it's basically just more background :)