From: Jerry Sievers <gsievers19@comcast.net>
To: Greg Stark <gsstark@mit.edu>
Cc: Robert Haas <robertmhaas@gmail.com>,  pgsql-hackers@postgresql.org
Subject: Re: time-delayed standbys
References: <BANLkTi==TTzHDqWzwJDjmOf__8YuA7L1jw@mail.gmail.com>
	<BANLkTi=QScpDTkUw3iw6JaO15uUvx_P4qg@mail.gmail.com>
Date: Wed, 20 Apr 2011 08:49:10 -0400
In-Reply-To: <BANLkTi=QScpDTkUw3iw6JaO15uUvx_P4qg@mail.gmail.com> (Greg
	Stark's message of "Wed, 20 Apr 2011 12:24:25 +0100")
Message-ID: <87wriphyux.fsf@comcast.net>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Greg Stark <gsstark@mit.edu> writes:

> On Wed, Apr 20, 2011 at 3:47 AM, Robert Haas <robertmhaas@gmail.com> wrote:
>
>> While I was out at the MySQL conference last week, I heard that one of
>> the forthcoming MySQL features is "time-delayed replication":
>
> Incidentally, this is a popular Oracle feature. It's a poor man's
> "flashback" and similar to how some filesystems automatically create
> regular snapshots of every home directory so you can get back stuff at
> some arbitrary point in the past.

Yup.

One of the big bosses where I work went asking for this a couple years
ago.  We're multi-platform; Oracle, MySQL, EnterpriseDB, Vertica.

They put a 6 hour delay on the critical Oracle boxes.  NOthing was done
for MySQL or Vertica since no feature support.

My C-foo being rusty, I elected to add more hacks to our home-spun
PYthon version of pg_standby rather than adding the feature to
pg_standby itself.

Been running with delayed WAL apply ever since..  Shipping happens
immediatly on log creation at master and we simply wait for the files to
age the configured time before application.

In a few cases, we have 2 or more standbys off the same prod master.
One in real-time mode and the others lagged.

Thanks all!

-- 
Jerry Sievers
Postgres DBA/Development Consulting
e: gsievers19@comcast.net
p: 305.321.1144