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

On Tue, 19 Aug 2003 12:04:11 +0200, Tom Ivar Helbekkmo <tih+nr_at_eunetnorge.no> wrote:

> Dieter Nöth <dnoeth_at_gmx.de> writes:
>
>> Please point me to the PostgreSQL documentation where [MVCC] is
>> descibed.
>
> In the manual, in this section:
>
> <http://www.postgresql.org/docs/7.3/interactive/mvcc.html>
>
> Locking is avoided by using MVCC, and reserving the option of failing
> a transaction (and raising an exception) if another transaction
> commits conflicting data during the attempted transaction. Thus, as
> Guido says, there are no locks, and nothing blocks anything -- but the
> application had better be prepared to retry a failed transaction.
>
> Of course, explicit locking is available, so you can run transactions
> the normal way, and never notice MVCC except as good performance. :-)
>
> -tih

Strikes me it's all a load of hokum, especially since that documentation goes on to say that if there's an index on the table, then lo and behold (index) page-level locks are taken (not that I'm surprised) as the index is updated to reflect changes on the underlying table. And if the index is locked, where does that leave the table?

HJR Received on Tue Aug 19 2003 - 05:15:39 CDT