Re: Heavy contgnous load

From: Jeff Janes <jeff(dot)janes(at)gmail(dot)com>
To: kzsolt <kzsoltkzsolt(at)freemail(dot)hu>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: Heavy contgnous load
Date: 2011-10-20 15:55:39
Message-ID: CAMkU=1yXGNzyFdAwzMVjJ11-P=Y-pF1cqjc2F1CtX7zRwrCyxQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On Tue, Oct 18, 2011 at 5:09 AM, kzsolt <kzsoltkzsolt(at)freemail(dot)hu> wrote:
> Dear Anybody!
>
> I use pgr to store records. But the characterisitc of the record traffic are
> special. For example 50 of them arrived in one sec contignously trough weeks
> and aligned interally trough tables.

What happens if the database has a hiccup and can't accept records for
a few seconds or minutes? Do the processes that insert the records
just buffer them up, or drop them, or crash?

> To absorb this traffic I put the pgr database to ramdisk (fast as possible).
> But after more day work the pgr slowing down.
> What is important think for this task I do not need any tranasction. So the
> COMMIT and ROLLBACK feature is useless.
> The question is how I minimize the rollback activity to free resoureces?

If you really don't need transactions, then you are incurring an awful
lot of overhead by using a transactional database.

In any case, this seem like a case for synchronous_commit=off. If the
database crashes, you might be missing a few seconds of recent
transaction when it comes back up. But, if the records are still
being generated while the database is down, you are losing those ones
anyway, so losing a few more retroactively may not be a big deal.

Cheers,

Jeff

In response to

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message kzsolt 2011-10-20 20:10:11 Re: Heavy contgnous load
Previous Message Bort, Paul 2011-10-20 13:24:45 Re: delete/recreate indexes