Hi, On 10/20/2015 04:33 PM, Scott Marlowe wrote:
We're running LSI MegaRAIDs at work with 10 SSD RAID-5 arrays, and we can get ~5k to 7k tps on a -s 10000 pgbench with the write cache on. When we turn the write cache off, we get 15k to 20k tps. This is on a 120GB pgbench db that fits in memory, so it's all writes.
I'm not really surprised that the performance increased so much, as the SSDs have large amounts of DRAM on them - with 10 devices it may easily be 10GB (compared to 1 or 2GB, which is common on RAID controllers). So the write cache on the controller may be a bottleneck.
But the question is how disabling the write cache (on the controller) affects reliability of the whole RAID array.
The write cache is there not only because it improves performance, but also because it protects against some failure modes - you're mentioned RAID-5 which is vulnerable to "write hole" problem.
regards -- Tomas Vondra http://www.2ndQuadrant.com PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
