Hi,
I am investigating the possibilites to upscale with MySQL. We've looked into Cassandra and MongoDB but the read performance is too low compared to MySQL so we are now looking to do it all in MySQL.
We have a myisam table in a db which has 500 partitions by key (max rows: 50M so 100.000 rows per partition). 100.000 rows is about 1Gb of data including indexes.
This in a single node is no problem. No for our challenge:
The reads on this table are about 1,000 per second...BUT the writes are about the same!!
We want to replicate the database using a master-master setup (=ndb cluster) using up to 10+ nodes. A master with 10 slaves is not good with so much writes so we hope master-master replication is.
My questions:
- is this setup even possible?
- is there a better solution (we need to keep the read performance very very high and be able to update/write on the same without disturbing the read performance)
Thx,
Maarten
I am investigating the possibilites to upscale with MySQL. We've looked into Cassandra and MongoDB but the read performance is too low compared to MySQL so we are now looking to do it all in MySQL.
We have a myisam table in a db which has 500 partitions by key (max rows: 50M so 100.000 rows per partition). 100.000 rows is about 1Gb of data including indexes.
This in a single node is no problem. No for our challenge:
The reads on this table are about 1,000 per second...BUT the writes are about the same!!
We want to replicate the database using a master-master setup (=ndb cluster) using up to 10+ nodes. A master with 10 slaves is not good with so much writes so we hope master-master replication is.
My questions:
- is this setup even possible?
- is there a better solution (we need to keep the read performance very very high and be able to update/write on the same without disturbing the read performance)
Thx,
Maarten