Embracing Web Services

Hope, an architect of the .NET enterprise server team, has worked on both the COM and XML teams in Microsoft. This experience allowed him to contrast the adoption of those two technologies within Microsoft. While COM took eight years to fully integrate across the product range, XML had permeated Microsoft within fourteen months!

This deep adoption of XML can be seen in Microsoft's plans for integration with .NET: SOAP, XML Schemas, UDDI -- laying down the foundation for cross-platform interoperability.

Hope identified what he saw as the key challenges for the new web services oriented architectures:

• An agreed n-tier programming model
• Management of metadata (e.g., from database schemas to interface specifications, where the newly invented WSDL solves some problems)
• Transaction models that deal with unreliability. Hope mentioned the compensation model here: rather than rollback models of transactions, there exist further actions that undo a transaction).
• Management of personal state data: personal data such as email, favorites, etc. is required anywhere at any time. There are some emerging solutions ("Passport" style) for this on the web.

Hope referred to the progress Microsoft has made with its XML technologies, highlighting the recently shipped MSXML 3.0. He said that other product teams within Microsoft are now dependent on that component, which has made sure that its quality and performance have had a lot of work. He also said that MSXML 4.0 could be expected in 2001, with support for XML Schemas and also XML Query as and when it is delivered by the W3C Query working group.

Concluding his presentation, Hope demonstrated the just-released beta of Visual Studio .NET. Microsoft have substantially re-written their IDE using their new .NET framework. Hope used it to show how easy building web services could be, taking a few steps to create a simple service using the C# language, accompanied by the automatic generation of the necessary interface definitions. (It was a little ironic that Hope's simple SOAP example was an RPC method, as he had taken pains to reiterate the SOAP-is-not-RPC mantra earlier.)

While the IDE certainly looked impressive and full of promise, some audience members adopted a "wait and see" attitude, saying that although it looked easy to create simple services, they wanted to see how it would stand up to complex systems. Other elements of potential interest in the .NET framework where XML is concerned included dual XPath/relational access to XML data.

The employment of XML at Microsoft certainly looks very encouraging. They have now brought XML pervasively through their product range, and are making more noises about cross-platform integration. As their .NET strategy unfolds we'll see how they deliver on that promise.