test_fsync output improvement

From: Bruce Momjian <bruce(at)momjian(dot)us>
To: PostgreSQL-development <pgsql-hackers(at)postgreSQL(dot)org>
Subject: test_fsync output improvement
Date: 2010-07-04 02:44:09
Message-ID: 201007040244.o642i9N07887@momjian.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

I have modified src/tools/test_fsync so its output is operations per
second, instead of total seconds, e.g.:

Loops = 10000

Simple write:
8k write 52549.436/second

Compare file sync methods using one write:
(open_datasync unavailable)
open_sync 8k write 575.599/second
(fdatasync unavailable)
8k write, fsync 349.669/second

Compare file sync methods using two writes:
(open_datasync unavailable)
2 open_sync 8k writes 215.472/second
(fdatasync unavailable)
8k write, 8k write, fsync 566.493/second

Compare open_sync with different sizes:
open_sync 16k write 247.207/second
2 open_sync 8k writes 219.824/second

Test if fsync on non-write file descriptor is honored:
(If the times are similar, fsync() can sync data written
on a different descriptor.)
8k write, fsync, close 719.469/second
8k write, close, fsync 1457.288/second

This should make the values more useful.

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

+ None of us is going to be here forever. +

Browse pgsql-hackers by date

  From Date Subject
Next Message Bruce Momjian 2010-07-04 03:12:56 Re: Needs Suggestion
Previous Message Robert Haas 2010-07-04 02:37:28 Re: reassign owned to change the ownership for op class and family