As we know, the programming model for using Windows Azure Table Storage is like the pic below
Every table can be partition base on custom Partition Key.
If we take the advantage of the Partition Key, we can easily create a multitenancy data struture.
For example, we want to create a multi user blog, just like blogger, we might have a table, call it "Posts", to store the post written by the users.
Obviously, we can use the username as Partition Key.
So base on the username, we can easily retrieve the corresponding set of data.
This kind of approach, data level multitenacy architecture, should be exist long time ago, but apply them onto cloud storage, will gain benefit that we cannot have in relational database.
Cloud storage will guarantee all data be stored highly distributed, and fully replicated.