db optimization

Hi,
I have a dual Xeon 3Ghz machine with 4Gb RAM 2x RAID1 for system and logs
and RAID10 for databases.
The machine has bad read performance and i am trying to look at things like:
controller config and how caching is setup for example on one of the machines
we have:
128Mb
Currently set to options (no ratios mentioned)
read Adaptive Read-ahead read-ahead or no read-ahead
write Write-back write-through
cache direct I/O cache I/O
It seems that no configuration changes were made to SCSI controller and
machines are in the state provided by DELL - same machine as above:
internal
Perc 3/di
Channel 1 2x 36Gb Raid 1 C&F partitions
Channel 1 2x 72Gb raid 1 E partition
external
Perc 3/qc
Channel 1 4x 72Gb Raid 1+0 D partition
Channel 1 2 x 72 Gb Raid 1 G partition
which shows that no channels are utilised.
If I cannot make immediate changes to hardware what trace should I run with
profiler for index tuning?
Currently the DB is reorg and reindexed to have 10% free space by the look
of it every sunday...
Any help appreciated.
George
Did you already isolate the bad read performance based on the perf mon read
throughput?
Are you using /3GB in boot.ini to allow SQL Server to use 2.7gig of memory?
You can use Profiler to find out what's taking the most time and see if
index is being properly created.
"George" <George@.discussions.microsoft.com> wrote in message
news:4AFF7740-5C8F-44D1-98CC-
> I have a dual Xeon 3Ghz machine with 4Gb RAM 2x RAID1 for system and logs
> and RAID10 for databases.
> The machine has bad read performance and i am trying to look at things
> like:
> controller config and how caching is setup for example on one of the
> machines
> we have:
> 128Mb
> Currently set to options (no ratios mentioned)
> read Adaptive Read-ahead read-ahead or no read-ahead
> write Write-back write-through
> cache direct I/O cache I/O
> It seems that no configuration changes were made to SCSI controller and
> machines are in the state provided by DELL - same machine as above:
> internal
> Perc 3/di
> Channel 1 2x 36Gb Raid 1 C&F partitions
> Channel 1 2x 72Gb raid 1 E partition
> external
> Perc 3/qc
> Channel 1 4x 72Gb Raid 1+0 D partition
> Channel 1 2 x 72 Gb Raid 1 G partition
> which shows that no channels are utilised.
|||[boot loader]
timeout=30
default=multi(0)disk(0)rdisk(0)partition(2)\WINDOW S
[operating systems]
multi(0)disk(0)rdisk(0)partition(2)\WINDOWS="Windo ws Server 2003,
Enterprise" /fastdetect
this is what's in the file currently, the server is windows 2003 Enterprise
with SQL 2000 Enterprise SP3
shall I just add /3gb switch or the /PAE one?
George
"Peter Yao" wrote:

> Did you already isolate the bad read performance based on the perf mon read
> throughput?
> Are you using /3GB in boot.ini to allow SQL Server to use 2.7gig of memory?
> You can use Profiler to find out what's taking the most time and see if
> index is being properly created.
> "George" <George@.discussions.microsoft.com> wrote in message
> news:4AFF7740-5C8F-44D1-98CC-
>
>
|||/3GB
"George" wrote in message
> shall I just add /3gb switch or the /PAE one?