Re: Toast, Text, blob bytea Huh?

From: Jan Wieck <JanWieck(at)Yahoo(dot)com>
To: Peter T Mount <peter(at)retep(dot)org(dot)uk>
Cc: Joe Conway <joseph(dot)conway(at)home(dot)com>, jason(dot)ory(at)ndchealth(dot)com, pgsql-hackers(at)postgresql(dot)org
Subject: Re: Toast, Text, blob bytea Huh?
Date: 2001-08-24 19:00:19
Message-ID: 200108241900.f7OJ0KU02513@jupiter.us.greatbridge.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Peter T Mount wrote:
> Quoting Joe Conway <joseph(dot)conway(at)home(dot)com>:
>
> > TEXT is a datatype which stores character data of unspecified length (up
> > to
> > the max value of a 4 byte integer in length, although I've seen
> > comments
> > indicating that the practical limit is closer to 1 GB -- not sure why).
>
> It may be something to do with the 1Gb splitting of the physical files
> representing a table... Unless it changed recently, a table was split over
> multiple files at the 1Gb mark.

No, it's because the upper two bits of the variable size
field are used as flags.

But in practice there are other limits that force you to keep
the objects you throw into text or bytea fields alot smaller.
When your INSERT query is received, parsed, planned and a
heap tuple created, there are at least four copies of that
object in the backends memory. How much virtual memory does
your OS support for one single process?

And by the way, TOAST is not only used for character data
types. All variable size data types in the base system are
toastable. Well, arrays might be considered sort of pop-tarts
here, but anyway, they get toasted.

Jan

--

#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #

_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Bruce Momjian 2001-08-24 19:19:28 Re: User locks code
Previous Message Stephan Szabo 2001-08-24 18:50:05 Re: PATCH proposed with new features for CREATE TABLE