Providing data services for SOA

Apart from IBM, which likes to refer to ‘information services’, the other vendors that compete with IBM’s Information Server tend to talk about ‘data services’ so lets stick with that as a moniker. As far as I can tell there are three of these competitors: Metamatrix, which was the first to announce a product, earlier this year; and Sybase and Composite Software, which came to market at or around the same time as IBM—that is, in the last 6 weeks or so. Since the Sybase Data Integration Suite is discussed in a separate article I will here concentrate on Composite Software, whose product coincidentally is also called Information Server, though Composite’s software of this name is now in version 4.0 rather than a first release as in the case of IBM.

The Composite Information Server (CIS) has historically been focused on enterprise information integration (EII). That is, supporting real-time queries running against multiple, heterogeneous front-end and back-end data sources. Arguably, the company is the leading pure player provider of such solutions, not least because CIS is embedded within both Cognos 8 and Informatica’s PowerCenter, which gives the company sales traction that it could not otherwise achieve on its own.

With version 4.0 of CIS, Composite has moved beyond the conventional EII environment and into support for SOA. Of course, everybody nowadays wants to support SOA but in this case it is not simply a matter of using web services internally but also to enable SOA to extend into the data layer. This is important because while conventional SOA resolves the spaghetti that most application architectures look like it is equally valid to attempt to unravel the Gordian knot that represents data architectures, and to do the two in concert: this is what data services is all about. Note, however, the distinction between transactional services and data services—the former is process-driven (and thereby falls easily within conventional approaches to SOA) whereas data services tend to be query driven, which requires a different approach—for example, transactional services are typically push-based while data services are pull-based.

In practice, the new features in version 4.0 have been largely customer driven (the company is particularly strong within financial services) and it includes significantly extended XML capabilities (along with support for XPath), especially with respect to accessing XML from within SQL statements; improved and extended caching (for better performance—needed especially to protect the environment from too many SOA access procedures); enhanced metadata capabilities including metadata search and support for third party modelling tools such as ER/Studio and ERWin; and additional security support. Other notable features include the ability to automatically generate WSDL (web service description language) as well as database views (which themselves can now be published into a schema).

Put simply, Composite’s view is that most data requirements within SOA are about data access: and that’s what CIS does. Unlike IBM’s Information Server, Composite’s product of the same name just aims to resolve that one issue. In other words, Composite is aiming to provide a best-of-breed solution for that single thing as opposed to IBM’s one-stop shop offering. Which approach is to be preferred is an old chestnut but there is certainly scope for both within the market.