Metadata tagging and data repostoreis are being integrated into SOA tool vendor products at a rapid pace. The premise for why they’re doing this is that today’s SOA landscape consists of many reusable Web Services that can’t be easily found, and once located, require a lot more information to use than what’s available in something like a raw WSDL.
The companies that are adding metadata repositories include BEA Systems with their acqusition of the FlashLine metadata repository. webMethods acquired Cerebra for embedding into their Fabric product. And IBM has announced that they plan to develop the Service Registry and Repository component for WebSphere.
One idea given for the addition of metadata repostories is that software reuse of services is being roadblocked because service components are not easily discoverable. While Web Service discovery is probably a Use Case that is sure to become more important as adoption of SOA picks up, and for companies that have adopted large-scale SOA this may be an issue, but finding out what services are available is usually not the problem. Today most companies using SOA don’t have a list of services like saleforce.com’s AppExchange. Most companies are more often than not still in the pilot-program stage, but that is sure to change as more and more Web-Service-based applications are developed.
The value today of metadata repositories is to better document usage and system relationships of the services. More often than not web services are being ’discovered’ today from comments and discussions that come up during meetings or in email, and when word gets out that XYZ services are being used on project A, then project B people want to be able to use those services too. But how? And who has permission to access them? And are there examples for how to use them? And what kind of dependencies exist between them? Where is the data coming from and how real-time is it? How can users/developers be notified if the interface or internal service logic of a service is changed or updated?
Providing detailed usage information and governance policies for the services offers great value. WSDLs were not designed to hold that kind of information and the UDDI concept hasn’t really caught on. As SOA deployments grow in size and complexity tools like those offered by Metadata repositories will be ready.
