Re: Question: amount of data in 1 read constrained by os-limit?

With your setup you will probably be limited to 8 blocks, i.e. 64K. To prove this, try the following from SQL*Plus:

alter session set events '10046 trace name context forever, level 8'; select something from large_unindexed table;

        (to force a tablescan).

In the resulting trace file you will find WAITs for

        db file sequential read
with

	p1 = file number
	p2 = starting block number
	p3 = number of blocks requested by oracle in read.

You should find p3 = 8 quite often.

However, if you restart the database with the init.ora parameter:

        use_readv = true

you may find that you can actually push your maximum up to 32 blocks per read.         

Jeroen van Sluisdam <Jeroen.van.Sluisdam_at_vrijuit.nl> wrote in article <35E66BC2.9E5913B_at_vrijuit.nl>...
> Hi,
>
> Probably this question is more appropiate for a Unix-group, but
> hopefully
> somebody here knows the answer. I have an Oracle db with a 8K
block-size
>
> and a multi_block_read_count of 32 so this would be 256K of data
that
> can be read in a single read. My problem is how do I determine if
really
>
> 256K can be read or my OS constrains this to a certain amount and
which
> amount
> this is (think about unix block sizes etc..)
>
> Data: HP-UX 10.01
> Oracle 7.3.2.3
> Multi-block_read_count 32
> Block_size 8K
>
> Thanks,
>
> Jeroen
>
>
Received on Fri Aug 28 1998 - 04:18:29 CDT