Re: ASM Record Deletions
- From: "Alex Gorbachev" <gorbyx@xxxxxxxxx>
- To: kevinc@xxxxxxxxxxxxx
- Date: Wed, 14 Jun 2006 23:33:21 +0200
As a DBA I wouldn't trust some magic tool (be it ASM v 10.0 or SAN box
or clever FS) trying to fix application design and implementation
errors. I stick to the rule that any additional intelligence layer
without deterministic behavior compromises availability of my system.
ASM assumes that your design is good (and so does RAC, for example)
but Oracle doesn't mention it much. Instead, it builds the picture of
ASM and RAC as a silver bullet to all our problems. Reality is that
nowadays companies prefer to save on proper design/implementation
hoping that smart software will fix all oddities as vendors promise.
Of course, hardware/software expenses can go up but hey, isn't people
resources are most expensive those days? :-) I think that this trend
on the contrary will keep us in the job! Oracle really cares about us
DBAs and our jobs! ;-)
2006/6/14, Kevin Closson <kevinc@xxxxxxxxxxxxx>:
Consider the fact that the majority of OLTP accesses hit the
minority of the database blocks (usually on the order of 80% hit
some 20% of the data) ...and that minority of blocks
moves. If you have, say, 50 disks and you add 10 and go through a
rebalance, will the hot blocks get extra representation on the
new spindles? No. Should they? Yes. Who is implementing stuff like that?
Not Oracle. Guess who does?
--
Best regards,
Alex Gorbachev
http://blog.oracloid.com
--
//www.freelists.org/webpage/oracle-l
Other related posts:
