Cluster / Sharding / Reporting

[Leon de Beer]

Hi,

I've been asked to architect a high volume mysql based system. Data integrety and scalability are key. About 2 years ago I had spend some serious time trying to use mysql cluster and encountered various issues with performance, migration which made me move away from it. Since then I met up with a MySQL DB "guru" who confirmed the same and suggested that a much better scalability approach would be to deploy a sharded infrastructure. I have 2 questions:

1) Has MySQL-Cluster undergone significant changes in the last few years which should make me re-evaluate my position?

2) If Sharding is a better way (I obviously accept the redundancy complications this brings which I will deal with through master / master replication) - the question I have is around reporting. I never figured out if e.g. I have 20 users on server (a) and 30 on server (b) how do I run a report / query which gives me a combined view of 50 users.

Thanks in advance

Leon