scaling MySQL in 3d sarah novotny – open databases and LAMP services net
things you’ve heard about scale write 1 / read many partitioning / sharding multimaster / rings memcached / nosql net
①large datasets ②high volume reads ③high volume writes net
storage choices engine options storage engine filesystem volume manager hardware net
①large datasets large objects many rows net
storage flexibility, reliability, clone-ability net
net
②high volume reads more memory fast disks more memory net
net myisam vs innodb
net not to be obvious, but -- read less data! compress data (if you can) don’t use limit
use thoughtful primary keys net
a short diversion to swap or not to swap that is the question net
③high volume reads writes more memory fast disks more memory fast disks net
③high volume writes choose your filesystem well understand how your filesystem and raid controller work together tune them to work in concert net
facebook game case: highly concurrent writes low risk of -- omg, i lost my most recent score! net
shard data innodb_log_flush_at_trx_commit=0 benchmark i/o schedulers net
free tools innotop maatkit MySQL proxy monitoring/trending cacti templates $monitoring_server – the one you know net
additional resources irc.freenode.org – #mysql – #maatkit mysql.com HPM2e - Baron Schwartz, Peter Zaitsev, Vadim Tkachenko, and Jeremy Zawodny net
credits swap image – ent/m10106/latest/ ent/m10106/latest/ special thanks to gabriel cain and mike hamrick for suggestions on content and slides net
Blue Gecko and contact info senk on #mysql net Blue Gecko provides Remote DBA services for companies around the world 7x24x365 support including monitoring, performance analysis, proactive maintenance and architectural guidance for small and large datasets.