Comments on: Tuning Latch Contention: Cache-buffers-chain latches

By: Sachin

Sachin — Thu, 27 Nov 2008 06:32:29 +0000

great doc to know about CBC.
1.I would like to know more about algorithm to access a buffer cache.
2.Please put the script code for waits10g.

thanks for wonderful explanation ..

regards,
Sachin

By: Aman....

Aman.... — Sat, 23 Aug 2008 13:44:36 +0000

Riyaj,
That was a great read and I guessed the table ORDER too :). I got a small doubt,what is latch children?
Regards
Aman….

By: Log Buffer #108: A Carnival of the Vanities for DBAs

Log Buffer #108: A Carnival of the Vanities for DBAs — Fri, 01 Aug 2008 17:04:18 +0000

[…] Pythian’s Riyaj Shamsudeen delved into tuning latch contention: cache-buffers-chain latches. […]

By: Riyaj Shamsudeen

Riyaj Shamsudeen — Thu, 31 Jul 2008 14:32:13 +0000

Hi yas

Thanks for your reply.

In this specific case, you are correct as we could get data_object_id from x$bh. But, this buffer could be an undo segment block, an undo header block or a free list bitmap block and x$bh.obj is set to high value (4294967295) for these type of blocks.

That’s why I prefer to dump the block and read through trace file. I Agree that there is more than one method to identify attributes of a block.

Cheers
Riyaj

By: Yas

Yas — Thu, 31 Jul 2008 08:01:56 +0000

Riyaj, as we know it is an expensive operation (latch-wise) to query dba_extents you can get the data_object_id of the object from x$bh.obj. There is no need to go to dba_extents.

By: Don Seiler

Don Seiler — Wed, 30 Jul 2008 19:00:53 +0000

Great post, Riyaj! It will still take me a couple more reads to hopefully understand it all.

By: Polarski Bernard

Polarski Bernard — Mon, 28 Jul 2008 06:19:49 +0000

Good work.

By: Scott Jenkins

Scott Jenkins — Sat, 26 Jul 2008 02:51:07 +0000

I now, I know why I leave the tuning to you. I still enjoyed the read.