From: | Markus Wanner <markus(at)bluegap(dot)ch> |
---|---|
To: | Simon Riggs <simon(at)2ndquadrant(dot)com> |
Cc: | Robert Hodges <robert(dot)hodges(at)continuent(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Bruce Momjian <bruce(at)momjian(dot)us>, Jens-Wolfhard Schicke <drahflow(at)gmx(dot)de>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Transaction-controlled robustness for replication |
Date: | 2008-08-13 13:38:32 |
Message-ID: | 48A2E3D8.5020601@bluegap.ch |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi,
Simon Riggs wrote:
> Classification of Replication Techniques
Thanks for your classifications. It helps a great deal to clarify.
> Type 2 is where you ship the WAL (efficient) then use it to reconstruct
> SQL (flexible) and then apply that to other nodes. It is somewhat harder
> than type 1, but requires less infrastructure (IMHO). Definitely
> requires less data shipping from Primary node, so very possibly more
> efficient.
What leads you to that conclusion? AFAICT a logical format, specifically
designed for replication is quite certainly more compact than the WAL
(assuming that's what you mean by "less data").
The only efficiency gain I can see compared to type 1 is, that most of
the processing work is offloaded from the master to the slave(s). For
setups with multiple slaves, that's a bad trade-off, IMO.
> Previously, most RDBMS vendors supported type 1a) systems. They have now
> moved to type 2 and 3 systems. Both DB2 and Oracle support a type 2
> *and* a type 3 replication system. The reasons they do this are valid
> for us also, so I suggest that we do the same. So for me, it is not
> about whether we do type 2 or type 3, I think we should do both.
I currently don't think type 2 is doable with any reasonable effort, but
hey, I'm always open for surprises. :-)
Which of IBM's and Oracle's products are you referring to?
Regards
Markus Wanner
From | Date | Subject | |
---|---|---|---|
Next Message | Stephen R. van den Berg | 2008-08-13 13:46:41 | Re: Replay attack of query cancel |
Previous Message | Magnus Hagander | 2008-08-13 13:34:55 | Re: temporary statistics option at initdb time |