The service oriented architecture (SOA) is often presented as the biggest architectural infrastructural and organisational project of modern IT. Many end user organisations, and most suppliers, are completely redesigning their software so that almost every function or item of data can be delivered as a loosely coupled service.
The promise of the service oriented world is dramatically increased business agility, and a far greater degree of fine-grained control over IT, which over the years has come to be viewed as too inflexible and too expensive.
This vision of a flexible, loosely-coupled, Lego-like world has a long history in IT. It can be traced back to the object-oriented and componentisation pioneers of the late 1980s and early 1990s, and was then developed through various iterations of web services initiatives before and after the dot-com boom and bust.
Since then, SOA has become much more serious, precisely defined, widely discussed, and, critically, widely implemented. There is no longer any debate about what SOA is, or whether the industry is moving in that direction: rather, it is about how best to implement a service oriented architecture, the management and technology issues, and whether the technology itself is mature enough. Above all, there is the question: Will it deliver on the hugely hyped promise?
Over the past year, the rapid adoption of the service oriented approach to IT means that answers
to some of these long-standing questions are now emerging. But new issues have also arisen, not least the impact of service oriented approaches on the underlying architecture, issues of governance and policy, and the relationship between new “software as a service” functions and SOA.
SOA: The technology
The fundamental and revolutionary idea of SOA is that the functions and business processes embedded in software can be easily packaged up and exposed to other programs as ‘services’ to which other programs can ‘subscribe’. These services can be effectively separated from the bigger, perhaps monolithic application in which they may be contained, and then ‘loosely-coupled’ with other systems to create end-to-end processes in support of the business. In this way, customers can effectively create their own new “composite” applications by reusing all the functions from elsewhere.
Even over the past year, much has been learned about the technology that is required if an enterprise class service oriented architecture is to be adopted. Among the key components are:
• An “integration and orchestration platform”. This is the system that is used to handle how all the different services interoperate with each other to ensure performance, scalability and security, and to provide a platform for composite applications and for services such as analytics.
• A “producer platform”. In order to gain the full business benefits of the SOA, businesses need to be able to carry out three distinct functions: model business processes for later deployment; build new applications or services; and create ‘composite’ applications using existing applications. In order to do this, they need specialist development tools.
• Specialist tools to wrap up legacy applications as services, create directories, and apply security.
• Service oriented business applications (SOBA). Many functions traditionally embedded in ‘monolithic applications’ are now being exposed as services. It is increasingly possible to compose and decompose packaged applications using third party tools.
• Further SOA “infrastructure” tools. These may be included in the integration engine, but may be excluded. They include much neglected but all-important functions, such as policy and service management, security, governance and mediation of services.
• A flexible underlying infrastructure (see article page 18). Demand for applications delivered as services may be less predictable than non-service but less flexible systems. They may be switched on or off, coupled or de-coupled at will. Ideally, a flexible, underlying IT infrastructure using, for example, virtualisation, will reduce or eliminate capacity problems.
For more on SOA, the technology and best practice, visit the Information Age SOA and Web services briefing room.
Encouraging signs
Given the sustained level of expect-ation and hype, it seems that SOA in business has been a long time coming. But it is coming.
Various surveys over the past year, including many from leading analyst firms and one by Information Age, concur that among bigger organisations (above SME level), about half of all businesses have a plan in place to implement a service oriented architecture in the near future.
This may be slightly high. The latest Information Age Effective IT survey suggests half are using web services for integration, but only 30% have taken this to an architectural, enterprise level. Either way, the adoption figures will rise steadily over the next five years until the approach is entirely mainstream.
The survey produced a more telling result: 92.5% of those businesses taking an SOA approach have reaped clear business benefits, defined as improved service levels, or saved money – an overwhelming endorsement. Thirteen per cent (13%) said it had done both, and was therefore classed as “very effective”. Just 7.5% thought they had not benefited from employing SOA.
The significance of this result should not be understated: enterprise wide SOA adoption is a big project, even if it is possible to start small, and it has cultural, organisational and technical implications. It is also expensive, so any reported favourable financial return at this stage is promising.
Lessons learnt
With so many businesses now adopting SOA, a body of expertise is also building up. At the beginning of the SOA project there were dire warnings that, alongside the successes, many businesses would suffer huge failures. If this is happening, it is not on a large scale and it isn‘t hitting the watchful IT business press (although big problems at some companies have come to light).
Some of the lessons that Information Age has heard over the past year include :
• The difference between success and failure is down to planning, using the right technology, and taking a business view from the start.
• Business alignment is critical for successful SOA deployment. It is important to take a business perspective at the top level, and in the way each service is viewed and delivered.
• Governance is often overlooked. Because it is so easy to create services, it can also be easy to lose control. Establish clear policies and standards for security and operational quality from the outset.
• SOA can pay for itself. The business benefits can be accrued along the way and re-invested.
• Small projects should be undertaken with an architectural and business end goal in mind. They should go ahead without waiting for full infrastructural roll out.
• Investment in services will be more readily recouped if they are shared with partners. Let them cross the firewall.
To date, the experience of SOA from most businesses has been more positive than experienced observers of the IT world might have expected.
For a long time, there has been near universal agreement that the SOA will ultimately prove to be a powerful and flexible way to organise and manage modern software. The report card on SOA so far is that, while care must be taken, more needs to be done and much potential is unfulfilled, it is not disappointing.
