LVM and VxVM

Fielsystem creation and Disk I/O performance

 
siva0123
Trusted Contributor

Fielsystem creation and Disk I/O performance

Hi,

I've got three disks allocated from the San as follows:

/dev/rdsk/c31t0d7 400 GB
/dev/rdsk/c31t0d6 60 GB
/dev/rdsk/c31t0d5 40 GB

I need to create some 25 filesystems spanning across 3 databases with the following sizes

DB1 => 192 Gb
DB2 => 169 Gb
DB3 => 34 Gb

I'm planning to create two vgs as follows :

vg01 ==> By combining disks /dev/rdsk/c31t0d7 and /dev/rdsk/c31t0d5 and create the filesystems
for the DB1 and DB2 in it

Vg02 ==> with /dev/rdsk/c31t0d6 and create the filesystems for DB3 in it.

All i want to know is will this plan have any Disk performance issues or is there any other better way of
configuration to have best performance.

Thanks,
Siva
6 REPLIES 6
Coolmar
Esteemed Contributor

Re: Fielsystem creation and Disk I/O performance

I believe you have to plan the disks carefully (seperate controllers and disks for indexes and databases, etc. With the SAN and fibre channel technology now, disk performance is rarely an issue in configuring database filesystems.
avik
Valued Contributor

Re: Fielsystem creation and Disk I/O performance

what exactly you mean when you say "I need to create some 25 filesystems spanning across 3 databases with the following sizes".

btw, these luns are made from the same raid group/disk group ?
Samir Pujara_1
Frequent Advisor

Re: Fielsystem creation and Disk I/O performance

Hi Siva,

First thing is it all depends upon the Storage Box and the LUN assigned from that.

Other thing is - it is always good idea to distribute all your filesystems with high transfer rate to multiple disk to get the optimum performance.

Samir
RAC_1
Honored Contributor

Re: Fielsystem creation and Disk I/O performance

First speak to SAN admin and ask him is there any strippping he did on SAN side. You don;t want to do stripping again if he has done something on san side. It will kill the performance.
There is no substitute to HARDWORK
Yogeeraj_1
Honored Contributor

Re: Fielsystem creation and Disk I/O performance

hi siva,

have a look at these documents:
http://www.oracle.com/technology/deploy/availability/pdf/OOW2000_same_ppt.pdf
http://otn.oracle.com/deploy/performance/pdf/opt_storage_conf.pdf

They talk about S.A.M.E - Stripe and Mirror Everything (SAME)

or you may wish to implement ASM (Automatic Storage Management) which implements MASE (mirrored stripes)

With ASM, you will just create one area and all the 3 databases will share it! The great thing is that the day you decide to add a new disk, you just have to add it to the ASM configuration and it is automatically available to all the 3 databases...


hope this helps!

kind regards
yogeeraj
No person was ever honoured for what he received. Honour has been the reward for what he gave (clavin coolidge)
siva0123
Trusted Contributor

Re: Fielsystem creation and Disk I/O performance

Hi ,

Thanks all for your valuable suggestions.


Siva