"Scott Marlowe" wrote: > I second this. Partitioning in time in past reporting databases > resulted in huge performance improvements for select queries.
Most statements will load data from a single year, but multiple monthes. I have a integer field containing the year and will use it for partitionning. It will also help a lot to remove one year after 4 years of activity. Actually this same database is used with 2 millions of lines per year (instead of 30). It is loaded with 3 years and runs quite fast unpartitionned on a 4 years old single SCSI HD with 2Go of RAM, single core pentium4. It runs a LOT faster on a quad xeon 2.83GHz with 8Go of ram and SATA HD, which is quite common now for dedicated servers. I tried partitionning on it: it showed no performance gain for such a small size, but it is an evidence that it will help with 30 millions of lines/year. OK, thanks to all your recommandations, I will ask hosters for a RAID10 4x250go SATA. -- Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
