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

Re: Rollback in Postgres

From: Kaare Rasmussen <kaare(at)jasonic(dot)dk>
To: pgsql-sql(at)postgresql(dot)org
Subject: Re: Rollback in Postgres
Date: 2008-07-14 20:34:33
Message-ID: 200807142234.33528.kaare@jasonic.dk (view raw or flat)
Thread:
Lists: pgsql-sql
> which "it would come in handy" wouldn't have enabled it.  (FWIW this
> feature used to exist in the Berkeley code, under the cool name "time
> travel", and was removed a long time ago.)

No, it didn't AFAIK. Timetravel kept all tuples in the database with all 
indexes and constraints active at all time. That's not the case with the 
flashback technology. You put aside some storage space that you don't need 
for something else. When that space is spent, tuples start dropping off the 
edge.

I've talked to people who was very much happy with this feature. Mostly DBA's 
recovering from their own stupid mistakes of course :-)

But yes, it has to be enabled, and yes it has to have a performance cost 
somehow, but people are requesting it, and somehow I don't think Oracle 
developed the feature just for fun. If you plug into Postgres' vacuum it 
would be rather cheap to make, I recon. I wouldn't worry about query speed as 
I guess that the use cases for retrieving already deleted rows don't aren't 
performance dependant.

-- 

Med venlig hilsen
Kaare Rasmussen, Jasonic

Jasonic                 Telefon: +45 3816 2582
Nordre Fasanvej 12
2000 Frederiksberg      Email: kaare(at)jasonic(dot)dk

In response to

Responses

pgsql-sql by date

Next:From: Kaare RasmussenDate: 2008-07-14 20:38:54
Subject: Re: Rollback in Postgres
Previous:From: Simon RiggsDate: 2008-07-14 20:20:38
Subject: Re: Rollback in Postgres

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