Re: Fine-grained replication?

Поиск
Список
Период
Сортировка
От Paul Jones
Тема Re: Fine-grained replication?
Дата
Msg-id 1340125827.62503.YahooMailNeo@web45503.mail.sp1.yahoo.com
обсуждение исходный текст
Ответ на Re: Fine-grained replication?  (Sergey Konoplev <sergey.konoplev@postgresql-consulting.com>)
Список pgsql-general
I should have been more specific about why I'm interested in this.

The idea is not so much the replication but being able to use the
replication information to make business decisions based on what is
observed happening to the data, or being able to selectively repair
data if there are problems.

XReader certainly shows a lot of promise, but it's not quite here yet.

Specific answers to respondents below...

> From: Sergey Konoplev <sergey.konoplev@postgresql-consulting.com>
> Date: Tue, 19 Jun 2012 13:00:49 +0400

> On Tue, Jun 19, 2012 at 1:49 AM, Paul Jones <pbj@cmicdo.com> wrote:
>> Is anyone aware of other non-trigger-based, fine-grained replication tools for PostgreSQL along the lines of the
XReader<http://wiki.postgresql.org/wiki/XReader> or pgreplay <http://pgreplay.projects.postgresql.org/>? 

> Binary Replication? http://wiki.postgresql.org/wiki/Binary_Replication_Tutorial
> Sergey Konoplev

Thanks!  This is good info.

On Tue, Jun 19, 2012 at 12:34:02PM +0200, Albe Laurenz wrote:

> I'm not sure what you want, because "pgreplay" is certainly not a
> replication
> tool, unless "replication" means something quite different to you than
> it
> does to me.
>
> Can you be more specific?

Yes, as explained above.  So pgreplay in this context would be more likely
to be used to repair something.

> Date: Tue, 19 Jun 2012 12:09:28 +0200
> From: Guillaume Lelarge <guillaume@lelarge.info>
>
> If you want fine grained replication, it means you won't be able to use
> log shipping and streaming replication. If you don't want trigger based
> replication, you won't have many options still available. pgPool comes to
> mind but, AFAIK, it isn't fine grained (and, to be honest, I would say it
> isn't really replication).
>
> So, nope, sorry.
>
> BTW, what's the issue with trigger-based replication? it really helps a
> lot in many cases.
>

The worry with trigger-based replication is that it could become a
performance bottleneck for heavy loads.  We tried out Slony just to see
how it worked and it did its job just fine for what we did with it.

Paul Jones

В списке pgsql-general по дате отправления:

Предыдущее
От: Benedict Holland
Дата:
Сообщение: Re: Smaller multiple tables or one large table?
Следующее
От: Scott Marlowe
Дата:
Сообщение: Re: Feature discussion: Should syntax errors abort a transaction?