Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Avoiding any locks in SQL Servers - read and understand....its magic.

Re: Avoiding any locks in SQL Servers - read and understand....its magic.

From: Tom Ivar Helbekkmo <tih+nr_at_eunetnorge.no>
Date: Tue, 19 Aug 2003 13:13:46 +0200
Message-ID: <86u18d528l.fsf@athene.i.eunet.no>


"Volker Hetzer" <volker.hetzer_at_ieee.org> writes:

> But shouldn't the one that *started* first win?

Nope. The one that commits first must win, otherwise it will be forced to wait until all other transactions, started after it, that can possibly be suspected of conflicting with it at some point, have completed. Every time a transaction completes, a check is made to see if it conflicts with a non-committed transaction at isolation level "serializable", and, if so, said transaction fails.

> Now, in your system,

Ahem. Not "my" system. SQL 1992. The international standard.

> transactions aren't isolated because a commit in another transaction
> can abort my own transaction evan if I "was there first".

Correct. Then again, the other transaction "was there first" in the sense that it got its change into place first, right?

Anyway, I'm not arguing for or against anything: I just pointed out that what's implemented in a few RDBMSes, including PostgreSQL, is in accordance with SQL 1992.

-tih

-- 
Tom Ivar Helbekkmo, Senior System Administrator, EUnet Norway
www.eunet.no  T: +47-22092958 M: +47-93013940 F: +47-22092901
Received on Tue Aug 19 2003 - 06:13:46 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US