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: Multi-Block read count

Re: Multi-Block read count

From: Bass Chorng <bchorng_at_yahoo.com>
Date: 9 Jan 2002 16:24:57 -0800
Message-ID: <bd9a9a76.0201091624.b3f1047@posting.google.com>


Connor McDonald <connor_mcdonald_at_yahoo.com> wrote in message news:<3C3C9F0E.38A6_at_yahoo.com>...

>
> For example, Oracle on Solaris can get up to 1m, but unless you set
> maxphys kernel parameter you won't get near that.
>

I think there is another important layer in between which also restricts your milti block count - that is your Filesystem cache, unless you use raw or Veritas quick IO.

I believe Solaris UFS has limitation of 8k cache IO size, for VxFS the default is 256K but you can uptune it to 1M.

My production box has maxphys set to 8m, ( I guess for Solaris 2.7, it only goes up to 1m regardless what you set it to ), but my block_size * multi-block will not exceed 256K regardless what I set my multi-block to. We are running VxFS with default setting.

I may be wrong on these theories because I obtain these thru my own test and investigation + some readings. Your corrections welcome. Received on Wed Jan 09 2002 - 18:24:57 CST

Original text of this message

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