Skip site navigation (1) Skip section navigation (2)

AW: 8192 BLCKSZ ?

From: Zeugswetter Andreas SB <ZeugswetterA(at)wien(dot)spardat(dot)at>
To: "'Christopher Kings-Lynne'" <chriskl(at)familyhealth(dot)com(dot)au>, Hackers List <pgsql-hackers(at)postgresql(dot)org>
Subject: AW: 8192 BLCKSZ ?
Date: 2000-11-28 09:49:09
Message-ID: 11C1E6749A55D411A9670001FA68796336814A@sdexcsrv1.f000.d0188.sd.spardat.at (view raw or flat)
Thread:
Lists: pgsql-hackers
> I don't believe it's a performance issue, I believe it's that 
> writes to
> blocks greater than 8k cannot be guaranteed 'atomic' by the operating
> system.  Hence, 32k blocks would break the transactions system.  (Or
> something like that - am I correct?)

First, 8k are not atomic eighter. Second, the page layout in PostgreSQL has been
designed to not care about the atomicity of IO. This design might have been 
compromised for index pages recently, to optimize index performance, 
but data pages are perfectly safe.

Andreas

pgsql-hackers by date

Next:From: Zeugswetter Andreas SBDate: 2000-11-28 09:51:28
Subject: AW: 8192 BLCKSZ ?
Previous:From: Hannu KrosingDate: 2000-11-28 09:27:51
Subject: Re: Full text Indexing -out of contrib and into main..

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group