Re: Postgresql replication

From: Chris Browne <cbbrowne(at)acm(dot)org>
To: pgsql-general(at)postgresql(dot)org
Subject: Re: Postgresql replication
Date: 2005-08-25 21:39:39
Message-ID: 60mzn5n1b8.fsf@dba2.int.libertyrms.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

steve(at)blighty(dot)com (Steve Atkins) writes:
> On Thu, Aug 25, 2005 at 12:20:52PM -0400, Chris Browne wrote:
>> bohdan(dot)linda(at)seznam(dot)cz (Bohdan Linda) writes:
>> > I would have a slight offtopic question, this is issue only of pgsql or
>> > there are some other db solutions which have good performance when doing
>> > this kind of replication across the world.
>>
>> Asynchronous multimaster replication is pretty much a generally tough
>> problem.
>>
>> Oracle, Sybase, and DB2 all have methods of doing it; none are either
>> simple, straightforward, or transparent to use.
>>
>> It's a tough problem, in general.
>
> I've been involved in one project that did multi-master async replication
> across geographically separated servers.
>
> It worked but was operationally so horrible that I've sworn never to
> architect a system that requires multi-master replication at the
> database ever again. Application level? Possibly. But not at the DB
> level.

I have heard something of the same argument being thrown at attempts
to use things like CORBA to allow you to pretend that there is no
difference between local and remote access to things.

There are some good arguments to be made, there.

You really need to access remote data in different ways than local
data because the latency *will* kill you if you ignore it.

It's a well and neat idea to try to find ways to hide those
differences under the carpet; it is quite likely that the "better way"
will involve addressing that somewhere in the application, as opposed
to trying to get the DB (or the ORB or other 'distributed thingie') to
do it for you.

Tables that contain balances (e.g. - credit balances, account
balances, and such) would be a good example. You do NOT want to
distribute the process of updating balances across some
hideous-latency "Great Divide" link. That heads, of course, to
application design, not to "pure DB level" activity...
--
(format nil "~S(at)~S" "cbbrowne" "cbbrowne.com")
http://www.ntlug.org/~cbbrowne/advocacy.html
"Tools that are no good require more skill."

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Lee Harr 2005-08-25 21:41:30 Re: drop table before create
Previous Message Ralph Mason 2005-08-25 21:37:39 Odd Problems