Re: Help me convincing IBM please!!!!
2005-04-07 14:06:56
On Thu, Apr 07, 2005 at 01:40:08PM -0400, David Longo wrote:
> I've been following this thread all day. Without running the script,
> I know I am getting abround 2.5 million objects per hour examined
> and it takes aboput 2.5 hours to run expiration daily. That's this
> week. Last week, it took about 8-10 hours to run expiration.
>
> What's the difference? Last week my DB was on a FAStT200HA
> with 4 drawers of 73GB 10K disks. This week it is on a
> FAStT 600 Turbo with 4 drawers of 73GB 15K disks and the EXP710
> drawers. That is the only change, Hint Hint! The TSM Buffer pool
> stayed the same size and my "Cache hit" pct was still 99% + in both
> cases. These disks systems are totally owned by one TSM server.
expiration is very I/O database dependent, and I agree with the 'hint' above.
I will provide my numbers below, but want to note that some time ago (about 3
months), I moved my database from 10KRPM 9GB drives, to 15KRPM 18GB drives
all logical volumes, TSM mirrored. My expiration performance moved up
dramatically! I do see a case for 'smaller' is better here, as long as you
don't saturate your controllers as well (I have always had plenty).
>From 03/08 - 03/29 I was running on a IBM F50 with 4 332MHZ procs and 3GB
memory. Cache hit stays around 99% (I allow TSM to manage the buffer pool).
On 3/30, I moved wholesale to a P5-570 with 4 1.6GHZ processors and 8GB memory.
Same U320 controller set up on the database and log disks as before.
I am using the same 15KRPM 18 GB drives for my database and log, all TSM
mirrored.
TSM server 5.1.9, AIX version 5.2 ML05
PCT_UTILIZED AVAIL_SPACE_MB
------------ --------------
85.3 69616
ACTIVITY Date Objects Examined Up/Hr
------------------ ---------- ---------------------------------
F50 -->
EXPIRATION 2005-03-08 2138400
EXPIRATION 2005-03-09 1864800
EXPIRATION 2005-03-10 1753200
EXPIRATION 2005-03-11 2109600
EXPIRATION 2005-03-12 2602800
EXPIRATION 2005-03-13 2545200
EXPIRATION 2005-03-14 2649600
EXPIRATION 2005-03-15 2278800
EXPIRATION 2005-03-16 2440800
EXPIRATION 2005-03-17 2019600
EXPIRATION 2005-03-18 2278800
EXPIRATION 2005-03-19 2134800
EXPIRATION 2005-03-20 1936800
EXPIRATION 2005-03-21 2599200
EXPIRATION 2005-03-22 2458800
EXPIRATION 2005-03-23 2635200
EXPIRATION 2005-03-24 2484000
EXPIRATION 2005-03-25 2206800
EXPIRATION 2005-03-26 2487600
EXPIRATION 2005-03-27 2206800
EXPIRATION 2005-03-28 1436400
EXPIRATION 2005-03-29 1360800
P5 -->
EXPIRATION 2005-03-30 1317600
EXPIRATION 2005-03-31 2041200
EXPIRATION 2005-04-01 1612800
EXPIRATION 2005-04-02 1580400
EXPIRATION 2005-04-03 1872000
EXPIRATION 2005-04-04 1875600
EXPIRATION 2005-04-05 1818000
EXPIRATION 2005-04-06 2048400
EXPIRATION 2005-04-07 1659600
I would be interested in knowing what people are using for their platforms,
and specifically what type of I/O infrastructure the database and log is on..
bob
|
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- Re: Help me convincing IBM please!!!!, (continued)
- Re: Help me convincing IBM please!!!!, Horst Scherzer
- Re: Help me convincing IBM please!!!!, Lawrence Clark
- Re: Help me convincing IBM please!!!!, Troy Frank
- Re: Help me convincing IBM please!!!!, Meadows, Andrew
- Re: Help me convincing IBM please!!!!, Julian Armendariz
- Re: Help me convincing IBM please!!!!, Dearman, Richard
- Re: Help me convincing IBM please!!!!, David Longo
- Re: Help me convincing IBM please!!!!, Jonathan Kaufman
- Re: Help me convincing IBM please!!!!, Andrew Ferris
- Re: Help me convincing IBM please!!!!, Lawrence Clark
- Re: Help me convincing IBM please!!!!, Warren, Matthew (Retail)
|
|
|