From: | John Mitchell <mitchelljj98(at)gmail(dot)com> |
---|---|
To: | Grzegorz Jaśkiewicz <gryzman(at)gmail(dot)com> |
Cc: | pgsql-general(at)postgresql(dot)org, John Mitchell <mitchelljj98(at)gmail(dot)com> |
Subject: | Re: What locking mechanism is used for database backup and restore and Master-Slave Replication? |
Date: | 2010-01-22 14:00:00 |
Message-ID: | fa2ac0581001220600n8d2a430xd6b3f822bf04989f@mail.gmail.com |
Views: | Whole Thread | Raw Message | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
When is the new version of postgres (8.5) scheduled to be released as the
latest stable version?
Thanks,
John
2010/1/21 Grzegorz Jaśkiewicz <gryzman(at)gmail(dot)com>
> depends on what sort of replication you are going to use really.
> Most are based on triggers. So they have a bit more impact on
> performance. As far as locking goes, postgresql is very conservative
> with locks, ie - it won't abuse them, unlike for instance mysql.
> So I don't know whether you are just worried about performance in
> general, or locking in particular. But impact of likes of slony isn't
> that big, but obviously once the new version is out - it will be even
> less of a problem, since postgresql already produces transactional log
> for recovering, and pretty much the same log is going to be used for
> replication.
>
>
> 2010/1/21 John Mitchell <mitchelljj98(at)gmail(dot)com>:
> > So am I to presume that the current stable version of postgres (before
> 8.5)
> > does require extra locking?
> >
> > John
> >
> > 2010/1/21 Grzegorz Jaśkiewicz <gryzman(at)gmail(dot)com>
> >>
> >> On Thu, Jan 21, 2010 at 1:12 PM, John Mitchell <mitchelljj98(at)gmail(dot)com>
> >> wrote:
> >> > Hi,
> >> >
> >> > In reading the documentation it states that the SQL dump backup does
> not
> >> > block other operations on the database while it is working.
> >> yes, pg_dump opens serializable transaction thus guarantees data to be
> >> the exact snapshot (as opposed to the default isolation level, which
> >> is called 'read commited' not without reason).
> >> >
> >> > I presume that while a restore is occurring that no reads or updates
> are
> >> > allowed against the restored database.
> >> nope, what restoring does, is just running all the commands in the
> >> pg_dump (whether it is binary or textual). So as soon as the database
> >> is created, it is treated just as any connection, thus allows you to
> >> connect and use it.
> >>
> >>
> >> > What locking mechanism is used for Master-Slave Replication?
> >>
> >> master slave that's introduced in what's to be 9.0 (aka 8.5), uses WAL
> >> shipping. So it doesn't require any extra locking.
> >>
> >>
> >>
> >> --
> >> GJ
> >
> >
> >
> > --
> > John J. Mitchell
> >
>
>
>
> --
> GJ
>
--
John J. Mitchell
From | Date | Subject | |
---|---|---|---|
Next Message | Alberto Colombo | 2010-01-22 14:39:44 | Re: timestamps, epoch and time zones |
Previous Message | Bruce Momjian | 2010-01-22 13:45:53 | Re: When is the release date for Postgres 8.5? |