ADSM-L

Re: Cache Hit Rate, Bufpoolsize Size and NT

2002-02-12 15:24:58
Subject: Re: Cache Hit Rate, Bufpoolsize Size and NT
From: "Prather, Wanda" <Wanda.Prather AT JHUAPL DOT EDU>
Date: Tue, 12 Feb 2002 15:22:23 -0500
Well, since no one else has ventured an answer, I'll take a SWAG in hope
that someone will at least provide more information in shooting it down!

My DB is 28 GB on AIX, and my buffpool is only 192,000; I agree that 327680
is QUITE high for a 2.5 GB, Windows or not.

You say that a consultant rebuilt your DB - I assume you are talking an
UNLOAD/RELOAD?
If so, then what I would do is - nothing, at least for a while.

Based on what little has been published about the structure of the TSM DB, I
know it is a b-tree.  When you reload a DB of that type, usually what
happens is that the DB becomes, shall we say "tightly packed", and a good
bit of DB activity takes place right after a DB reload as the tree splits
off new empty "leaves" to accomodate new incoming records.  If that is what
is happening here, then over time your DB cache hit should gradually get
better again, as the DB "resettles" itself into normal adding patterns.

What I would do is reset the cache statistics, watch it for a couple of
days; then reset them again, watch it for a couple days; see if it gets
better on its own after a week or two.

OK, that's the only far-fetched answer I can come up with.  Any body else
had strange numbers after a DB rebuild?
************************************************************************
Wanda Prather
The Johns Hopkins Applied Physics Lab
443-778-8769
wanda_prather AT jhuapl DOT edu

"Intelligence has much less practical application than you'd think" -
Scott Adams/Dilbert
************************************************************************






<Prev in Thread] Current Thread [Next in Thread>