You should be able to achieve what you described using Fan-out Querying for Federations in SQL Azure. There is a very nice (2 part) article, written by Cihan Biyikoglu, about this technique:
What is a fan-out query?
Federations provide a model for partitioning parts of your schema over to multiple member databases for harnessing scalability of many nodes. However applications still need for querying all of the data across federation members. Fan-out is a technique
for querying data in your federation, across many federation members. Fan-out queries are much like map/reduce in that it is formed in 2 parts;
Member query is the piece that is sent over to all members involved in the query and Summary query is the query that is the post processing piece to allow condensing the results from the member query to desired final result-set.
With fan-out queries the member query is always there but summary query may not be needed. For example if you are simply doing DML (we’ll have some examples like data pruning or reference data management etc) or DDL (we’ll look at schema deployment in detail
below), fan-out would only have a member query but no summary query is needed. It is only when you need post processing, you need the summary query.
i could be wrong, but it seems only using fan-out query utility page (http://federationsutility-scus.cloudapp.net/) to get results from all members ? I need to do it pragmatically, mean in my code. I can't use this utility page.
i also need do data aggregation like count, ave, max/min, like summary queries