Matthew wrote:
> On Fri, 25 Jan 2008, Greg Smith wrote:
>> If you're seeing <100TPS you should consider if it's because you're
>> limited by how fast WAL commits can make it to disk. If you really
>> want good insert performance, there is no substitute for getting a
>> disk controller with a good battery-backed cache to work around that.
>> You could just put the WAL xlog directory on a RAID-1 pair of disks to
>> accelerate that, you don't have to move the whole database to a new
>> controller.
>
> Hey, you *just* beat me to it.
>
> Yes, that's quite right. My suggestion was to move the whole thing, but
> Greg is correct - you only need to put the WAL on a cached disc system.
> That'd be quite a bit cheaper, I'd imagine.
>
> Another case of that small SSD drive being useful, I think.
PostgreSQL 8.3 will have "asynchronous commits" feature, which should
eliminate that bottleneck without new hardware, if you can accept the
loss of last few transaction commits in case of sudden power loss:
http://www.postgresql.org/docs/8.3/static/wal-async-commit.html
--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com
