|
Hi, I’m currently dealing with performance issues of
postgres and looking for some advice. Platform Postgres: 7.0.2 OS: FreeBSD4.4 DB: size - about 50M, most frequently updated tables are of
an average size of 1000-2000 rows and there are not many of them, about 15 in
total Description My current system load keeps the postgres CPU utilization at
the level of 90-100%. ‘vacuumdb’ results in a sharp drop of the CPU
usage down to 25-30%, but not for a long period of time – it gets back to
100% within 30 minutes. Disk IO ratio during the test keeps on about 0.5 MB/s Questions: 1. When reading the ‘vacuum analyze’ output how
to identify which one of the actions had the most effect on reducing the CPU
usage – garbage cleaning or statistics recalculation for the analyzer? 2. What would be the recommended set of parameters to tune
up in order to improve the performance over the time, instead of considering an
option to vacuum every 30 minutes or so? 3. Is it safe to run ‘vacuum’ as frequently as
every 15-30 minutes? 4. Suggestions? I know that 7.0.2 is an old version and therefore ran the
same test on 7.3.18 – the performance behavior was similar. Thank you in advance, Sergey _________________________________________________ This
message, including any attachments, is confidential and/or privileged and
contains information intended only for the person(s) named above. Any other
distribution, copying or disclosure is strictly prohibited. If you are not the
intended recipient or have received this message in error, please notify us
immediately by reply email and permanently delete the original transmission
from all of your systems and hard drives, including any attachments, without
making a copy. |
