Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Do I really need more than 1 rollback segment?

Re: Do I really need more than 1 rollback segment?

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Thu, 20 Dec 2001 09:46:57 -0000
Message-ID: <1008843095.21643.1.nnrp-12.9e984b29@news.demon.co.uk>

Nuno,

Your solution to the 1555 is correct -

Given that over time all rollback segments will grow to the same size (if optimal is not set, and there are no silly games with on/off line, and no silly numbers in extent sizes etc...), your rollback segment size has to be large enough to allow the single longest event that requires consistent reads to complete at its worst possible run time.

It is possible, of course, that your change from 25MB minimum size to 400MB is
an over-reaction that is causing redundant I/O to take place - but if you can't see any side-effects on the I/O subsystem. who cares ?

You MAY want to experiment with reducing the rollback segment size to see if this improves the I/O load on the system.

Alternatively you may decide that the rollback segments MUST be that size to avoid 1555 - in which case you MAY reduce the total
rollback segment volume by reducing the
number of rollback segments - at which point Howard's observation kicks in: if you have too few rollback segments you start to see contention for the rollback segment header block (the GETS reported in v$rollstat start colliding, so the WAITS count starts to climb).

--
Jonathan Lewis
http://www.jlcomp.demon.co.uk

Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

Author of:
Practical Oracle 8i: Building Efficient Databases

Screen saver or Life saver: http://www.ud.com
Use spare CPU to assist in cancer research.

Nuno Souto wrote in message <3c1738cd.1717929_at_news>...

>Howard J. Rogers doodled thusly:
>
>>
>>Personally, I go for the Steve Adams school of thought on this one. I
can't
>>see any drawbacks (though I'm sure Jonathan will elaborate on the
>>'additional I/O' idea) of large segments. It's the NUMBER of them that's
>>the worry, to avoid contention issues.
>>
>
>Well, this is interesting. I used to get a lot of 1555 errs with 16
>rlbsegs with 5M extents and 5 minextents. No optimal, shrink early
>morning and late arvo.
>
>I've changed it to 16 rlbsegs, 20M extents, 20 minextents. Again, no
>optimal, same shrinks.
>Not a single 1555 since then.
>Application is mix of OLTP and DSS (Peoplesoft Financials).
>
>Users are around 200, but actual connections are less than 50 (using
>an app server).
>
>
>Cheers
>Nuno Souto
>nsouto_at_optushome.com.au.nospam
Received on Thu Dec 20 2001 - 03:46:57 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US