On Wednesday, January 18, 2017 at 8:43:54 PM UTC+1, Troels Arvin wrote:
Hello,
When looking at db2look output for an old tablespace, I saw:
CREATE LARGE TABLESPACE ...
...
EXTENTSIZE 64
PREFETCHSIZE 96
...
OVERHEAD 7.500000
TRANSFERRATE 0.060000
DROPPED TABLE RECOVERY OFF
This got me thinking: When using a SAN storage system with automatic
tiering between SSD and various kinds of rotating disk, what should one
put for OVERHEAD and TRANSFERRATE? Does DB2 even care about those values nowadays?
I suppose EXTENTSIZE and PREFETCHSIZE could matter, but I would hope that
DB2 is somewhat smart about prefetching etc. Or?
I've turned dropped table recovery off because I believe I've once seen
it contribute to a state where a tablespace had trouble releasing free
space after a tablespace-full situation. But my memory could be wrong. At
any rate: I'm rather scared of restoring tables into a live tablespace
any way, so it's unlikely that I'll make use of DROPPED TABLE RECOVERY.
Does someone have experiences/opnions to share on this?
--
Regards,
Troels Arvin
Steve Rees just published an excellent article exactly about that:
http://www.idug.org/p/bl/et/blogaid=589
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)