Hi,
I wonder if someone has had any experience in finding a general
optimisation order in MySQL Cluster.
This document that has helped me on single machine MySQL 4.1 installs:
MySQL Presentations: Optimizing MySQL -
http://dev.mysql.com/tech-resources/presentations/presentation-oscon2000-20000719
<snippet>
Optimizing hardware for MySQL
* If you need big tables ( > 2G), you should consider using 64 bit
hardware like Alpha, Sparc or the upcoming IA64. As MySQL uses a lot
of 64 bit integers internally, 64 bit CPUs will give much better
performance.
* For large databases, the optimization order is normally RAM,
Fast disks, CPU power.
* More RAM can speed up key updates by keeping most of the used
key pages in RAM.
.........
</snippet>
I guess it still applies to Clusters, but probably not with Disk Access?
I realise YMMV, but say I had a theoretical cluster with:
250Mb Database. Fairly Simple Table Structure that was reasonably
well indexed (most queries take 0-1 seconds). SELECTs outnumber
UPDATE/INSERTs 5:1. The Cluster is hit with an average of a 3 queries
per second, but this is known to fluctuate +/- 500%
1 x Management Server (512Mb RAM, Pentium 4 2.8)
2 x Storage Nodes (1024Mb RAM, Pentium 4 2.8)
1 x API Node (1024Mb RAM, Pentium 4, 2.8).
All 10/100 Network. Each machine running Software RAID1 with 10k RPM
SATA drives.
At a glance would you upgrade the Network First, the CPUs or the
number of nodes? Would you not upgrade the CPUs but just add more
nodes?
P.S. Haven't yet used MySQL cluster in production environment..
replication seems to be working fine.
