> No idea what mysql thinks a shard is, but in PG we have read-only hot standby's.
I used sharding as an expression for partitioning data into several databases.
Each user in the system is unaware of any other user. The user never accesses the private data of another user. Each
usercould in theory be assigned their own database server. This makes it easy to split the 40000 users over a number of
databaseservers. There are some shared data that is stored in a special "shared" database.
> The standby database is exactly the same as the master (save a bit of
> data that has not been synced yet.) I assume you know this... but I'd
> really recommend trying out PG's hot-standby and make sure it works the
> way you need (because I bet its different than mysql's).
> Assuming the "shared" and the "sharded" databases are totally different
> (lets call them database a and c), with the PG setup you'd have database
> a on one computer, then one master with database b on it (where all
> writes go), then several hot-standby's mirroring database b (that
> support read-only queries).
As our data is easily partitioned into any number of servers we do not plan to use replication for load balancing. We
dohowever plan to use it to set up a backup site.
> Its pretty hard to guess what your usage pattern is (70% read,
> small columns, no big blobs (like photos), etc)... and even then we'd
> still have to guess.
It's more like 40% read 60% write.
> Not only will I not compare apples to oranges, but I really wont compare
> apples in Canada to oranges in Japan. :-)
Hehe
/Lars