Microsoft's 'Orleans' cloud programming model gets a Halo test drive

Remember Microsoft Research's Orleans cloud-programming project? The Halo team at Microsoft does.
Written by Mary Jo Foley, Senior Contributing Editor

Whenever anyone from Microsoft describes a Microsoft Research project, there's almost always a disclaimer -- specifically, "There's no guarantee when and if this technology will ever see the commercial light of day."


Given that caveat, it's interesting when a Microsoft Research project takes a step forward. Recently, that seems to have happened with "Orleans," a cloud-programming model I last blogged about back in 2010.

When perusing the end-of-year research roundups, I found this interesting tidbit buried in the eXtreme Computing Group's list of accomplishments for 2012:

"The cloud-systems team celebrated a year of successful deployment of its distributed cloud technology—Orleans—in production for Microsoft’s Halo team, and the team has scaled its system very significantly since then."

That's one pretty impressive proof-of-concept demo.

Orleans, as Microsoft officials themselves have described it, "offers a simple programming model build around grains, a unit of computation with private and shared state that communicates exclusively by sending messages to other grains and receiving and replying to requests from clients. Combined with the Orleans runtime, which provides functionality commonly used in this type of system, Orleans raises the level of abstraction and helps developers build scalable correct Cloud applications."

The (non-fiscal) Cliff Notes version: On the run-time front, Orleans is to Windows Azure approximately as the Common Language Runtime (CLR) is to Windows.

As the Microsoft researchers note on their page about Orleans: "Orleans provides direct support for the .NET programming model. We use standard .NET languages with custom attributes (currently C# is supported; we are working on F#)."

There's still no word as to when or how the eXtreme Computing Group will bring Orleans to market. Will it be internally used infrastructure only? Or will it be made available to any/all Windows Azure customers at some point? Stay tuned....

Editorial standards