Microsoft Oslo: Vaporware As A Service

Microsoft's announcement on Tuesday of "Oslo" sounded impressive:  a planned model-based development platform for SOA and BPM. But there also was some bad news for Microsoft's SOA ambitions. It's SOA-as-Service offering, which it calls an ISB (Internet Service Bus), probably won't be available before 2009.

Unlike Oslo itself, the ISB isn't pure vaporware. Microsoft has been letting people test parts of it at BizTalk Labs since July, and talks about making it into a commercially viable service. But this week, we learned that Oslo will be an integral part of BizTalk Services 1, Microsoft's current name for the offering. So SOA as a Service will have to wait for Oslo.

Just how long is that wait? Microsoft is being extremely coy about shipping dates, saying only that a beta will be available at some point during 2008. But it clearly wants us to believe that this means the final product (or service, in this case) will be out in 2009. As part of its launch press kit, it even sent out a (not entirely uncritical) report from analyst firm Directions on Microsoft that predicted the shipping date as 2009.

That may be too optimistic, especially as Microsoft is only implying 2009, not actually saying it. This article at The Register suggests that some components, notably Visual Studio 10, will not ship until 2011.

BizTalk Services probably isn't quite that far away: Oslo is a part of five different products, which Microsoft says will all have different release dates.  And there's no need for SaaS to have clearly defined releases or version numbers; one of its advantages is that bugs can be fixed and features added several times a day if necessary.

The ISB will probably be rolled out gradually, at first without Oslo. Microsoft might even be hoping it gains some traction while still in beta, though I can't see many customers going for that. Betas may be trendy in Web 2.0, but enterprise SOA users tend to want things a bit more stable.

Of course, BizTalk Services isn't just aimed at giant enterprises. Part of Microsoft's pitch for it is that SaaS opens up SOA and BPM to users who previously couldn't justify the expense of a dedicated ESB. However, this is happening anyway, thanks to commoditization and Microsoft's biggest threat, open source. And most SOAs link together internal systems, for which there's little reason to go to an outside service provider at all.

The real killer app for SOA as a Service is linking together different organizations (the original promise of Web services.) That's difficult right now, thanks to security and interoperability problems, but it gets much easier when both participants have their apps hosted by the same provider. The big question is who this provider will be. If Microsoft isn't careful, it could be Salesforce.com, Google, or even one of the Web 2.0 startups.