Enterprise SOA 2.0
From IRF 2008
[edit] Topic 1: Infrastructure for Enterprise SOA 2.0
In taking the step from intra- and inter-enterprise applications to the wider setting of the Internet of Services, extended capabilities are required from enterprise SOA. Enterprise SOA 2.0 as such should go beyond carefully controlled installation base and collaboration to “lighting” up the network with open global service supply and demand of enterprise services. Open service partnerships and ecosystems are crucial for this - for value-added service aggregation, delivery intermediaries and driving new market channels.
- What are the impediments for scaling Enterprise SOA from intra- and inter-enterprise applications, to the wider setting of the Internet of Services?
- Will software-as-a-service be the model of the future or can we expect other business models in a globalized flat world?
- What are new areas at the technical and business side that will require standardization?
Infrastructure for enterprise SOA 2.0 has to do with the broad notion of how will service-oriented architecture be adopted. There are many well established and understood technical challenges that must be adapted to the world of enterprise SOA. There are new challenges as well. There are two ways that SOA will be adopted. SOA will be adopted inside a company to help build solutions that will be used to create solutions for internal use. SOA will also be adopted to create public services, part of the so-called service grid that was analyzed in depth during IRF 2007. There is a large grey area between these two paradigms in which internally developed services are expose for use by the public or by partners or in which public servicies are used to help create internal solutions.
The main areas that must be addressed no matter what kind of adoption are:
- Governance: How to design processes for creating a collection of services, incorporating collections of third party services, and creating and enforcing rules for the use of services. Governance keeps the largely distributed process of adopting SOA from becoming a mess.
- Service Repository: How can you create a repository that describes all available services, shows services that have been proposed, and allows the use of services to be registered and SLAs to be presented and agreed to. A service repository should also provide an operational history of each service.
- Semantic descriptions and interoperability: Services descriptions are usually provided on the level of the interface to the service. It is also important the functions performed by the service, also called the semantics, are described. To avoid large amounts of translation from on format to another, services must use common terms and act on common data to be interoperable.
- Service Discovery: It must be possible to search for available services and find them.
- Service Level Agreements: It must be possible to enter into service level agreements that describe the performance the service provider will guaranteed that the usage levels expected from the service consumer as well as other terms.
- Operational Complexity: Composite applications built on services will each use many services. This means that the reliability and performance of one service might affect many applications. How can the operational characteristics of many services be tracked as well as the dependencies of applications on services and the dependencies of composite services.
- Scalability: As services become more popular they must be able to scale to handle the load.
- Standards: A variety of standards exist for web services and other aspects of SOA that must be adopted. New standards must be designed.
In addition to these technical concerns there is the broad challenge of semantic consistency and interoperability. In the IRF 2007 conference and book we examined these notions in detail. Some analysts, like John Seely Brown, see the emergence of service aggregators that guarantee quality and interoperability. It is also possible that instead of aggregation of a collection of third party services, a consistent grid may only be assembled by acquisition, that is the grid has a brand the same way that software products do now. The activities on the web today include both aggregation and consolidation.
