Re: PostgreSQL vs. InnoDB performance

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Peter Eisentraut <peter_e(at)gmx(dot)net>
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: PostgreSQL vs. InnoDB performance
Date: 2005-06-03 00:32:00
Message-ID: 29252.1117758720@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> On a particular system, loading 1 million rows (100 bytes, nothing
> fancy) into PostgreSQL one transaction at a time takes about 90
> minutes. Doing the same in MySQL/InnoDB takes about 3 minutes.

What sort of hardware, exactly?

Simple division says that that's about 11K transactions per minute,
which is more or less what you could expect to get with a 15000RPM
drive if everyone is honest and a commit actually involves bits hitting
a platter. Now we've talked about schemes for committing more than one
transaction per disk revolution, but there's no way we could get to 30
per revolution given our lack of knowledge about the actual disk layout.

I don't think I believe that InnoDB is really truly committing 330K
transactions per minute. Suggest that the customer try a pull-the-plug
type of test. Does the DB come back at all, and if so how close to the
last reported-committed row has it got?

regards, tom lane

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Michael Fuhr 2005-06-03 00:47:12 Re: SQL call to get pid of current connection
Previous Message Gevik babakhani 2005-06-03 00:30:46 PostgreSQL Developer Network