RE: Stupidity with ASM rebalancing

There are some improvements in 11g regarding rebalancing performance,
especially in the area of removing serialization events. I don't have
enough information about your environment, but if you are using
diskgroups with normal redundancy you may be hitting a known issue with
ASM (10g and 11gR1) that is excessive repartnering: ASM tries to have
only 10 partners per disk and this can 'cause extra work'. You can see
that by querying v$asm_operation, the column EST_WORK keeps growing as
the rebalancing goes on. BTW ASM disks are said partners when they have
copies of at least one mirrored extent.
If you want to know more about the origin of your performance problem
with rebalancing I suggest you try first to look at the arbx log files
in the BDUMP of the ASM instance and then possibly use a 10046 trace
against those processes.
If it can be of any help you can have a look at this recent presentation
of mine from UKOUG2007:

From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Finn Jorgensen
Sent: Saturday, December 08, 2007 2:21 AM
To: hrishys@xxxxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Subject: Re: Stupidity with ASM rebalancing

I had a 3TB diskgroup I added 2TB to and the rebalance took 3 days. Way
too long and I was very nervous the whole time it should break and I'd
have to start over which had happened before on smaller additions.
Not sure if this has been fixed in 11g. Haven't had the time to play
with it.

On Dec 7, 2007 5:22 AM, hrishy <hrishys@xxxxxxxxxxx> wrote:

        Everybody here says that there is some stupidity with
        ASM balancing i am just wundering what problems you 
        faced with ASM Disk reblancing and how has it been
        improved in 11g

        Sent from Yahoo! - the World's favourite mail <> 


Other related posts: