Oleg: I concur with this, and would like to collaborate as possible.
[You may recall my messages from a couple weeks back. I'm looking at
ways of implementing 'concurrent' PostGres on a NUMA machine that has
both local memory and (a bit more costly) access to a shared memory].
Bettina Kemme's method relies on fast, reliable and ordered global
communication, which is easily implemented on my target system, so her
methods look a great fit. [Im my case, the 'replication' would only happen
at the level of the buffer cache in each node, since all nodes have access
to a shared file system].
Presently, per Tom Lane's suggestion, I'm
looking at a different approach, in which I keep the shared memory segment
in shared memory and check-in/check-out the buffers into local memory at
proper times (which Tom suggested at the LockBuffer boundaries].
I'm open to suggestions/advice/...
cheers Mauricio
>From: Oleg Bartunov <oleg@sai.msu.su>
>To: <pgsql-hackers@postgresql.org>
>Subject: [HACKERS] PostgreSQL replication
>Date: Thu, 3 May 2001 10:37:09 +0300 (GMT)
>
>Hi,
>
>here is an answer from Bettina Kemme - author of
>Postgres-R - consisent replcation engine based on Postgres 6.4
>(URL of paper - http://citeseer.nj.nec.com/330257.html).
>While I don't see much activity on replication topic I think it's
>worth to discuss with Bettina design issue and co-ordinate
>our plans for future release.
>
> Regards,
> Oleg
>_____________________________________________________________
>Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
>Sternberg Astronomical Institute, Moscow University (Russia)
>Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
>phone: +007(095)939-16-83, +007(095)939-23-83
>
>---------- Forwarded message ----------
>Date: Wed, 02 May 2001 18:26:58 -0400
>From: Bettina Kemme <kemme@cs.mcgill.ca>
>To: Oleg Bartunov <oleg@sai.msu.su>
>Subject: Re: Postgres-R question
>
>Oleg,
>
>we are currently working on producing a Postgres-R
>version based on PostgreSQL 6.4 that can be released
>for the public. Darren Johnson from sourcesoft institute is
>actually working on this version right now.
>Since PostgreSQL 7.1 has a quite different concurrency
>control system than 6.4, upgrading Postgres-R to 7.1 is a bit more
>difficult.
>One of my students is going to work on this problem
>during the summer (he is going to start next week, but at
>the beginning he will have to do some reading to get familiar
>with replicated etc).
>We would be very happy to cooperate with the PostgreSQL
>developers and make the replication part a part of PostgreSQL
>(or at least an add-on feature).
>The problem is that our approach requires, to a certain degree,
>to change the underlying concurrency control method and
>the transaction execution control.
>That is, different to lazy replication schemes, it will
>not be exclusively an add-on feature, but must be integrated into
>existing code. Of course,
>we are planning to design the system such that the user
>(or system administrator) can choose between a replicated
>and a non-replicated system. In the non-replicated system,
>simply the original PostgreSQL will be executed.
>
>In the next days, I am going to have a closer look at
>PostgreSQL 7.1 to figure out all the details. Again,
>we are open for any kind of suggestions, support, cooperation
>etc.
>
> Best
> Bettina
>
>Oleg Bartunov wrote:
> >
> > Hello,
> >
> > I found your paper about database replication and your implementation
> > based on Postgres. What's the status of your implementation ?
> > Is't available for downloading ? I'm postgres developer and interested
> > in database replication. Current version of PostgreSQL is 7.1 and
> > it has much more feature than that of 6.4.2 but it still lacks
>replication.
> > I think your contribution would be very useful for other users.
> >
> > Regards,
> > Oleg
> > _____________________________________________________________
> > Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
> > Sternberg Astronomical Institute, Moscow University (Russia)
> > Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
> > phone: +007(095)939-16-83, +007(095)939-23-83
>
>--
>******************************************
>Bettina Kemme
>School of Computer Science
>McGill University Phone: +1 514 398 8930
>Montreal, Quebec Fax: +1 514 398 3883
>Canada E-mail: kemme@cs.mcgill.ca
>http://www.cs.mcgill.ca/~kemme
>
>
>---------------------------(end of broadcast)---------------------------
>TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com
