My understanding of parallel dump performance is that it only makes a difference when you have a large number of DBs (thousands if not tens of thousands). We performed similar testing using 9.3.x and found little performance gains using -j (with 100+ tables). See Bruce Momjian’s post : http://momjian.us/main/blogs/pgblog/2012.html Mel Llaguno • Staff Engineer – Team Lead Office: +1.403.264.9717 x3310 www.coverity.com <http://www.coverity.com/> • Twitter: @coverity On 5/4/15, 6:17 AM, "Jan Lentfer" <Jan.Lentfer@xxxxxx> wrote: >Am 2015-05-01 17:16, schrieb Susan K. McClure: > >[...] >> My postgresql.conf has these options for (hopefully) pg_dump and >> pg_restore improvements: >> >> ================= >> work_mem = 1GB # dump/restore Perf Value >> #maintenance_work_mem = 2048MB # min 1MB >> maintenance_work_mem = 1GB # dump/restore Perf Value >> #autovacuum_work_mem = -1 # min 1MB, or -1 to use >> maintenance_work_mem >> >> ... >> fsync = off # dump/restore Perf Valuecheckpoint_segments = 60 # >> dump/restore Perf Value >> checkpoint_warning = 60s # dump/restore Perf Value >> ...... >> autovacuum = off # dump/restore Perf Value.. >> ============================= >> >> Any thoughts on what else I might try to improve things ? >[...] > >It ist more interesting to know how your database schema is. How many >"large" tables do you have. What is e.g. the size of your 10,20,30 >largest tables? > >Regards > >Jan > > > > >-- >Sent via pgsql-admin mailing list (pgsql-admin@xxxxxxxxxxxxxx) >To make changes to your subscription: >http://www.postgresql.org/mailpref/pgsql-admin -- Sent via pgsql-admin mailing list (pgsql-admin@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-admin
