Re: Advice on partitioning

[Mehdi Salarkia]

Sorry for my late response because of the fact that we were having a lot of discussions and testing different solutions and testing them.

So after some more research and considering Ricks advises we decided to focus more on sharding and have the right clustered in index for our tables .
We are in the process of implementing the sharding and testing it but meanwhile we could deal with a big load of data without partitioning and by adding the right clustered index (which by the way strangely called primary key and MYSQL is not very clear about the difference between an index and primary key its documentations, or atleast I didn't see it) we could server 20,000 users with incoming/outgoing mails with a database of size 550,000,000 records.
We are also considering to make the table smaller and move some of the information to a clustered NO-SQL database (for now we are working on Riak and we are testing it) to store mail information and remove the mail heavy information from database.

Here is my last status update.
Hope that it will be useful for others, I'll come back to share more information a little latter.