Thoughts on Windows Server AppFabric, Windows Cluster and Deployment
So I’ve been watching some of the PDC 2009 videos. Muralidhar’s “Windows Server AppFabric Caching” session enlightens us with the fact that the notification mechanism now has a batch facility, primarily because XBox Live was getting 1.5m notifications per second. Sounds like transaction support is coming to Velocity – I guess this will leverage the STM.NET. The fact the the application polls the cache for notifications raise the question of how real-time the notification mechanism is?
Microsoft’s current roadmap from an application hosting perspective appears to be driven by IIS – Windows Server AppFabric is a set of integrated technologies that make it easier to build, scale and manage Web and composite applications that run on IIS. Unfortunately, not everything in the sell-side world of finance is driven through IIS – actually very little is. If you look at the sell-side Linux world it’s fairly common for Java processes to be run on a VCS cluster – these processes maybe listening to a message bus, calculating and pushing out new greeks or prices onto another bus. Looking at the Microsoft equivalent, Windows Cluster, it’s clear that AppFabic is Microsoft focus and Windows Cluster (MSCS) is legacy – just look at the .NET support (you have Generic Service support, but for the Failover Cluster APIs looks to be C++) and the lack of any PDC 2009 session on Windows Cluster.
Am I missing something on this whole issue of deployment?