Scott has published the results of his “Continuation Token” spike, which is a critical aspect of dealing with queries against Windows Azure tables storage. His findings will make it to the guide, but you can read the essentials here.

The unusual thing that you’ll see in his article is that it shows a way of dealing with forward and backward paging. The trick is storing the Continuation Token you get from Windows Azure in a stack (in session) and then using that to retrieve the right page of data. This is possible because the Continuation Token is serializable and you can persist it somewhere for later use.

There are some interesting implementation details I’d suggest you look at it if you have to deal with pagination.