test_fsync open_sync test

Is there a value to this test_fsync test?

Compare open_sync with different sizes:
(This is designed to compare the cost of one large
sync'ed write and two smaller sync'ed writes.)
open_sync 16k write 242.563 ops/sec
2 open_sync 8k writes 752.752 ops/sec

It compares the cost of doing larger vs. two smaller open_sync writes.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +

Bruce Momjian wrote:
> Is there a value to this test_fsync test?
>
> Compare open_sync with different sizes:
> (This is designed to compare the cost of one large
> sync'ed write and two smaller sync'ed writes.)
> open_sync 16k write 242.563 ops/sec
> 2 open_sync 8k writes 752.752 ops/sec
>
> It compares the cost of doing larger vs. two smaller open_sync writes.
>

Might be some value for determining things like what the optimal WAL
block size to use is. All these tests are kind of hard to use
effectively still, I'm not sure if it's time to start trimming tests yet
until we've made more progress on interpreting results first.

--
Greg Smith 2ndQuadrant US greg(at)2ndQuadrant(dot)com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us
"PostgreSQL 9.0 High Performance": http://www.2ndQuadrant.com/books

Greg Smith wrote:
> Bruce Momjian wrote:
> > Is there a value to this test_fsync test?
> >
> > Compare open_sync with different sizes:
> > (This is designed to compare the cost of one large
> > sync'ed write and two smaller sync'ed writes.)
> > open_sync 16k write 242.563 ops/sec
> > 2 open_sync 8k writes 752.752 ops/sec
> >
> > It compares the cost of doing larger vs. two smaller open_sync writes.
> >
>
> Might be some value for determining things like what the optimal WAL
> block size to use is. All these tests are kind of hard to use
> effectively still, I'm not sure if it's time to start trimming tests yet
> until we've made more progress on interpreting results first.

OK, thanks for the feedback. I just wanted to make sure it wasn't a
stupid test.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +

Greg Smith wrote:
> Bruce Momjian wrote:
> > Is there a value to this test_fsync test?
> >
> > Compare open_sync with different sizes:
> > (This is designed to compare the cost of one large
> > sync'ed write and two smaller sync'ed writes.)
> > open_sync 16k write 242.563 ops/sec
> > 2 open_sync 8k writes 752.752 ops/sec
> >
> > It compares the cost of doing larger vs. two smaller open_sync writes.
> >
>
> Might be some value for determining things like what the optimal WAL
> block size to use is. All these tests are kind of hard to use
> effectively still, I'm not sure if it's time to start trimming tests yet
> until we've made more progress on interpreting results first.

FYI, I beefed up the testing for this in test_fsync:

Compare open_sync with different write sizes:
(This is designed to compare the cost of writing 16k
in different write open_sync sizes.)
1 16k open_sync write 723.824 ops/sec
2 8k open_sync writes 347.979 ops/sec
4 4k open_sync writes 215.114 ops/sec
8 2k open_sync writes 174.613 ops/sec
16 1k open_sync writes 87.496 ops/sec

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +