Hi,

I am a newbie in db horizontal partitioning and I could see the potential of using this technology in the near future for my project. So allow me to ask a dump question. I don't really understand why we need a framework like Hibernate Shards, wouldn't most databases do these transparently these days? I did a quick search on this topic yesterday and found that mysql and postgresql support horizontal partitioning to a certain extent. It seems that mysql does this pretty well and my understanding is you don't have to create a shard-aware code for this as the database will do everything for you behind the scene. You can sort, group by, or even join regardless where the data reside (I know that Hibernate Shards doesn't support cross-shard relationships). But again I am new to this and would appreciate if someone can clarify if my assumptions above are correct or if I am missing something. Thanks in advance...

Thanks for posting, you've asked a good question. I can't speak for postgres, but I assume in the MySQL world you're referring to the horizontal partitioning features introduced in 5.1. From what I've read, this feature allows you to partition your data within a single db instance. That's great, and if your data fits in a single db and it's acceptable for all that data to live in a single physical location, you don't need Shards. However, if you simply have too much data or you're looking for geographic distribution of your data, Shards can help. The horizontal partitioning features available in 5.1 should help you get by with a single database for longer, but it still has its limits.

Hope that helps.

Max