So my graduate databases class recently took a look at several of these cloud computing environments, and Microsoft offering is clearly compelling both from an tool integration stand point and from the technical side.

For instance Microsoft's incremental innovation in their S3 like utility, called Simple Data Storage (SDS), is that SDS provides guarantees for Durability and Consistency, while Amazon really only guarantees Durability. There has been some work on implementing idempotent logs on top of S3, and SQS to provide similar guarantees but this is third party academic work see http://portal.acm.org/citation.cfm?id=1376616.1376645 for an introduction.

While at first glance Microsofts innovation is an improvement however, there is the question of how SDS performs at scale. Providing consistency for such a system will incur some amount of overhead. Has anyone here used SDS to serve larges amount of data with frequent updates? I would be interested to know how it compares with S3.