Re: improvement suggestions for performance design - Mailing list pgsql-performance

From Y Sidhu
Subject Re: improvement suggestions for performance design
Date
Msg-id b09064f30707050910x5924ed05w224b280c9a1e3f8f@mail.gmail.com
Whole thread Raw
In response to Re: improvement suggestions for performance design  (tfinneid@ifi.uio.no)
List pgsql-performance


On 7/5/07, tfinneid@ifi.uio.no <tfinneid@ifi.uio.no> wrote:
> On 7/5/07, tfinneid@ifi.uio.no < tfinneid@ifi.uio.no> wrote:
>
> I don't know much about this EAV stuff. Except to say that my company is
> in
> a situation with a lot of adds and bulk deletes and I wish the tables were
> designed with partitioning in mind. That is if you know how much, order of
> magnitude, data each table will hold or will pass through (add and
> delete),
> you may want to design the table with partitioning in mind. I have not
> done
> any partitioning so I cannot give you details but can tell you that mass
> deletes are a breeze because you just "drop" that part of the table. I
> think
> it is a sub table. And that alleviates table bloat and excessive
> vacuuming.

By partitioning, do you mean some sort of internal db table partitioning
scheme or just me dividing the data into different tables?

There want be many deletes, but there might of course be some.
Additionally, because of the
performance requirements, there wont be time to run vacuum in between the
insert, except for in non-operational periods. which will only be a couple
of hours during the day. So vacuum will have to be scheduled at those
times, instead of the normal intervals.



---------------------------(end of broadcast)---------------------------
TIP 7: You can help support the PostgreSQL project by donating at

                 http://www.postgresql.org/about/donate

Internal db table partitioning. Check out: http://www.postgresql.org/docs/8.2/static/ddl-partitioning.html

--
Yudhvir Singh Sidhu
408 375 3134 cell

pgsql-performance by date:

Previous
From: tfinneid@ifi.uio.no
Date:
Subject: Re: improvement suggestions for performance design
Next
From: smiley2211
Date:
Subject: Re: Query is taking 5 HOURS to Complete on 8.1 version