Re: ORACLE on Linux - IO bottleneck

Hello,

first of all, thanks for your answer...

> From what I read here: your system is asking for 512byte (wsec/wkB and
> rsec/rkB) read and writes.
> Physically the writes are done in 2k (w/wkB) while the read in less of
> 1k (7647,73/8151,98).
> The service time is low (0,11) probably due to the really short reads
> but it result in an inefficient operation.
>
> If the I/O is sequential you would benefit of an I/O scheduler...

I didnīt know what was an IO scheduler. Iīve been looking for and what Iīve seen is that kernel 2.6 implements one but kernel 2.4 doesnīt, isnīt it?

Iīve not understood very well the IO Scheduler function but Iīll read more about it ...

> > 1.- There is a generic low performance at the system, at SO level,
> > at oracle level and at application level. ŋCan we suppose that there
> > is an IO bottleneck?
>
> Possible but not sure.
> You are using your disks poorly, sorry, :(

Yea!, OF COURSE, this is the only thing Iīve got very clear and this is my main problem ...

> > 2.- ŋAre the "wrqm/s" / "rrqm/s" values correct?
> >
>
> No, they are low. You are not merging the I/O properly (maybe because
> your reads are not sequential).

OK. Reads should not be secuential in a OLTP database, Iīm wrong? So, I suppose reads are not secuential as you say ...

> > I've been reading different manuals and documentation about Raw Devices
> >
> > and Direct I/O in Linux (from Oracle an Redhat). Everything makes me
> > think that oracle
> > blocksize (8K) should not affect performance drastically because we use
> > DirectIO.
> >
>
> Direct I/O is not always the solution...

ĄNote it! Thanks...

> The storage statistics are right for writes. Check if the reads are done
> in about 1k.

Sure, reads are in about 1K.

> Just to be sure: check the readahead of your device.
>
> ( The /sys pseudo filesystem contains the information. Even hdparm can
> give you the number).

hdparm says readahead in that device (and in all other) is 120. I didnīt know what this parameter meant but Iīve been reading about and what Iīve understood is that for non-secuential reads readahead should be smaller than in a secuentially big file reading, isnīt it?

Thanks for all Fabrizio. Received on Thu Feb 09 2006 - 01:35:14 CST