ADSM-L

Re: query performance problem

2004-02-19 11:03:18
Subject: Re: query performance problem
From: Ervins Tumulkans <Ervins.Tumulkans AT VERDI DOT LV>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Thu, 19 Feb 2004 18:00:35 +0200
        I dont want to remove them i want to make  "move data" to relaim them 
to get fresh empty tape.
        The problem is that TSM not filling tapes which has for example 5% 
filled and had
        1% reclaim space because of expired data between them. So there is 100 
tapes 5%filled but has 2%reclaim space so I have 93% wasted space. So i need to 
select them and make move data to same storage pool to get empty.After that I 
can migrate whole 1TB disk pool with just with few tape mounts filling new 
empty tapes instead of 50 half filled tape mounts.  



> ----------
> 
> 
> From:         ADSM: Dist Stor Manager[SMTP:ADSM-L AT VM.MARIST DOT EDU] on 
> behalf of Thomas Denier[SMTP:Thomas.Denier AT MAIL.TJU DOT EDU]
> Sent:         Thursday, February 19, 2004 17:44
> To:   ADSM-L AT VM.MARIST DOT EDU
> Subject:      Re: query performance problem
> 
> > I'm trying to select tapes where has reclamation percentage and utilization
> perc
> > entage summ of less then 5. this is useful for reclamation as these tapes
> usless
> >  in library. My problem is my select return result after about 3-7minutes.
> Can a
> > nyone help to optimize this select or any tips
> > how to better use inner select.
> >
> > TSMserver version:5.1
> > Database size: 100GB
> > hardware: RS/6000 H80
> > RAM:2GB
> >
> >
> > SELECT="select
> volume_name,stgpool_name,pct_utilized,pct_utilized+pct_reclaim as
> >  summa
> > from volumes
> > where volume_name in (select volume_name from media where state like
> 'MOUNTABLEI
> > NLIB')
> > and stgpool_name like 'BACK1_3590'
> > and READ_ERRORS = 0
> > and WRITE_ERRORS = 0
> > and ERROR_STATE like 'No'
> > and STATUS not like 'EMPTY'
> > and ACCESS not like 'UNAVAILABLE'
> > and pct_utilized+pct_reclaim < 5
> > order by pct_utilized"
> >
> As far as I can tell, the 'like' and 'not like' operations could
> just as well be equality and inequality comparisons. The wild
> card capabilities of 'like' might have some overhead even when
> they are not really used.
> 
> I don't get the underlying goal. As far as I can tell, the script
> will find pending volumes and filling volumes with small amounts of
> data. I don't see why you would want to remove the filling volumes
> from the library. I can imagine circumstances where one would want
> to remove pending volumes, but those volumes could be found with a
> significantly simpler select.
> 

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