tuning postgresql writes to disk

From: Vinubalaji Gopal <vinubalaji(at)gmail(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: tuning postgresql writes to disk
Date: 2011-02-04 00:08:27
Message-ID: AANLkTinOPwGT+pg0bAHEL_geeHFqFipLEPr8_8fi+aj-@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Hi,
  I was wondering if Postgresql can be tuned such that it writes to
disk every n seconds or until the buffer reaches a certain threshold
to tune the performance . This is a special case where the application
wouldn't care if there is a data loss of 1 seconds or less. I would be
interested  to know if there is some configuration parameter which
already does this.  I looked at the WAL parameters and the new async
commit  but not sure if I am looking at the right place. Say i have 10
clients connecting and each client is inserting a record. I want to
force Postgresql only to write to disk when all the 10 records have
been written to the buffer or after 1 second (configurable). My
question is can I partly  achieve this by setting wal_writer_delay =
1000 ms or is there some other parameter which will help me to achieve
the desired effect?

--
Vinu

In a world without fences who needs Gates?

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Tom Lane 2011-02-04 02:54:07 Re: isn't "insert into where not exists" atomic?
Previous Message Mage 2011-02-03 23:59:45 Re: isn't "insert into where not exists" atomic?