At 10:00 AM 9/20/2005 -0400, Vivek Khera wrote:
On Sep 14, 2005, at 9:45 AM, vinita bansal wrote:
I have a 4 proc. AMD Opteron machine with 32 GB RAM and ~400GB HDD
and a 40GB database. I need to take backup of this database and
restore it some other location (say some test environment). I am
currently using pg_dump and pg_restore utilities to get this done
which takes 4-5 hrs for a dump and 8-9 hrs for restore
respectively. I am using custom format for taking dumps.
i'll bet you've saturated your disk I/O bandwidth, since for me
dumping a db a bit larger than that takes roughly 1 hour, and restore
about 4.
I don't think disk I/O is saturated, unless the database is very fragmented
on disk.
Most modern drives can manage at least 40MB/sec sequential throughput. Even
random seeks of 64KB or 128KB blocks should get you about 6-12MB/sec. So 4
hours is quite slow. And 8-9 hours for a restore of 40GB probably won't be
very pleasant if you have a boss or customer breathing down your neck...
Any reason why Postgresql would only get 2.8MB/sec for dumps or slower?
Regards,
Link.
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
