Re: Performance impact of record sizes

From: Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us>
To: John Moore <postgres(at)tinyvital(dot)com>
Cc: Postgresql Admin <pgsql-admin(at)postgresql(dot)org>
Subject: Re: Performance impact of record sizes
Date: 2002-07-04 19:47:28
Message-ID: 200207041947.g64JlSM04801@candle.pha.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

John Moore wrote:
> We have a need to store text data which typically is just a hundred or so
> bytes, but in some cases may extend to a few thousand. Our current field
> has a varchar of 1024, which is not large enough. Key data is fixed sized
> and much smaller in this same record.
>
> Our application is primarily transaction oriented, which means that records
> will normally be fetched via random access, not sequential scans.
>
> The question is: what size thresholds exist? I assume that there is a
> "page" size over which the record will be split into more than one. What is
> that size, and does the spill cost any more or less than I had split the
> record into two or more individual records in order to handle the same data?
>
> Obviously, the easiest thing for me to do is just set the varchar to
> something big (say - 10K) but I don't want to do this without understanding
> the OLTP performance impact.
>

If you don't want a limit, use TEXT. Long values are automatically
stored in TOAST tables to avoid performance problems with sequential
scans over long row values.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026

In response to

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Bruce Momjian 2002-07-04 19:50:28 Re: views: performance implication
Previous Message Gregor Mosheh 2002-07-04 19:25:09 memory strangeness