Just finished my last talk. Slides are downloadable here, and also embedded after the break.
MySQL Conf – Managing Terabytes
Own it: Working with a changing open source community
The floor show is closed, so no more booth work tomorrow. I’ve had a great time here talking with people and seeing my colleagues in the PostgreSQL and MySQL community.
Looking forward to getting some hacking time in tomorrow and enjoying an evening connecting with people instead of working on slides. 🙂
I’ve heard of databases with maybe one thousand tables and thought those bordered on the preposterous, but 409,994 tables is beyond anything I could imagine! How can one remain sane looking at a schema that size? Are many of the tables partitions of larger logical tables?
Hi!
The tables are a result of an inheritance structure that is essentially partitioned per customer. Each customer has about 12 tables that are created when their account is created.
Sanity is probably not possible, but we have a special tool written to view and document only the top level tables, and the many tables which are not part of the inheritance structure.
We’re hoping to open source that tool in the next few weeks.
-selena
Thanks, Selena. So I’m guessing each customer is in its own schema also? That (or some naming scheme by customer) would make the problem more tractable.
No, everything is in ‘public’.