Hi Kevin,
glad to be of help.
If you haven't seen it yet, you can setup schedules in the netapp cmd line
admin windows (maybe in a gui too, but I wouldn't know about that) for doing
reallocates.
I looked for a script I made to do this, but I guess I deleted it when I got off
the netapp.
For comparison, and future planning, my databases are on raid 10 arrays, with
15k 600 GB sas disks.
For a db on a 2 x 7 array, here are 5 dbb backup points. The array is san
attached.
I use 2 db streams, the output files are on sata raid.
Activity Start Time End Time Elapsed (hh:mm:ss)
Gigs
-------------- ------------ ------------ -------------------
----------
FULL_DBBACKUP 02-23-12.31 02-23-17.46 05:14:49
2338.08
FULL_DBBACKUP 03-02-09.38 03-02-13.02 03:24:30
2335.73
FULL_DBBACKUP 03-09-10.48 03-09-15.33 04:44:29
2437.75
FULL_DBBACKUP 03-16-11.11 03-16-14.46 03:35:18
2371.32
FULL_DBBACKUP 03-23-11.19 03-23-15.31 04:12:48
2363.04
And for a 2 X 3 array internal to a Dell r910 using 1 stream to sata raid -
Activity Start Time End Time Elapsed Min Gigs
-------------- ------------ ------------ ------------ ----------
FULL_DBBACKUP 02-23-11.00 02-23-12.40 100.6 817.73
FULL_DBBACKUP 03-02-11.00 03-02-12.25 85.9 772.45
FULL_DBBACKUP 03-09-11.00 03-09-12.27 87.6 774.07
FULL_DBBACKUP 03-16-11.00 03-16-12.33 93.0 789.22
FULL_DBBACKUP 03-24-13.53 03-24-15.21 88.2 795.65
Good Luck!
Bill Colwell
Draper Lab
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Kevin Kettner
Sent: Thursday, March 27, 2014 12:34 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: Moving the TSM DB2 database (AIX)
Thanks a ton for that info!
This is hugely promising. My volumes are definitely fragmented. I have 5
200 GB vols that make up my DB volumes on my worst server. Four of them
were recommended for reallocation. I started a reallocate on one last
night and it's 80% done. I can already see a huge improvement in my DB
backup rate. I expect it will slow down to my normal rates when it hits
the next volume, but I can already see it's helping:
DB backup rate graph
The DB backup rate started out nearly twice as fast today.
On 3/26/2014 13:04, Colwell, William F. wrote:
> When I had TSM databases on Netapp - both v5 & v6 - I had to do frequent
> netapp
> 'reallocate' commands to get the physical order in the netapp to match
> the logical order of db2.
>
> The db2 backup is reading the database sequentially, but within the netapp it
> is completely out of order.
>
> Try doing ' wafl scan measure_layout' to get a measure of the disorder.
>
> Bill Colwell
> Draper Lab
>
> -----Original Message-----
> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf
> Of Kevin Kettner
> Sent: Wednesday, March 26, 2014 1:24 PM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Re: Moving the TSM DB2 database (AIX)
>
> That is what I'm leaning towards as well. I've got NetApp looking at the
> disk end to see if its getting hit hard. I also want to find out if the
> work load is heavier on reads or write (I'm guessing read) to know what
> sort of hardware fix is best for this, cache, flash, or more spindles, etc.
>
> From Wanda's email, they're using a DS3512. I wouldn't expect that to
> be much different, performance wise, than the NetApp 3160 that I'm
> using. That leads me to think that maybe it's not the disk afterall...
>
> On 3/26/2014 11:30, Ehresman,David E. wrote:
>> Kevin,
>>
>> My gut reaction is that your disk drives can't feed the data fast enough.
>> If it were me, I would open up a PMR to find out what the real bottleneck is.
>>
>> David Ehresman
>> University of Louisville.
>>
>> -----Original Message-----
>> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf
>> Of Kevin Kettner
>> Sent: Wednesday, March 26, 2014 11:34 AM
>> To: ADSM-L AT VM.MARIST DOT EDU
>> Subject: Re: [ADSM-L] Moving the TSM DB2 database (AIX)
>>
>> On the original question, I have moved DBs on AIX with LVM several times
>> with good success. The only real concern is the performance impact. The
>> benefit over mirroring is you can do it with no outage at all.
>>
>> I have 3 servers with similar sized DBs on AIX with NetApp SAS disk on
>> the back end, backing up to IBM 3592 drives, and my DB backups take 4-6
>> hours. I'm on 6.3.4 now and I've tried using more streams but that has
>> not made much difference.
>>
>> My smallest production DB is around 200 GB and it takes about an hour to
>> backup.
>>
>> I wonder what's going wrong. Do you have any advice?
>>
>> Thanks!
>>
>> On 3/21/2014 15:42, Prather, Wanda wrote:
>>> Was just thinking the same -
>>> It's only the conversion from V5 to V6 that takes forever. Once you are
>>> V6/DB2, DB backup-restore is fast again.
>>>
>>> I have TSM 6.3.4 on Windows, DB is 930G used, DS3512 disk, and it will back
>>> up to LTO5 in 90 minutes if the server isn't doing a lot else at the time.
>>> Restore takes maybe 15 minutes longer.
>>>
>>> You've got other issues you should address, if your DB backup is taking
>>> many hours @ 300GB....
>>>
>>> W
>>>
>>>
>>> -----Original Message-----
>>> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On
>>> Behalf Of Ehresman,David E.
>>> Sent: Friday, March 21, 2014 9:11 AM
>>> To: ADSM-L AT VM.MARIST DOT EDU
>>> Subject: Re: [ADSM-L] Moving the TSM DB2 database (AIX)
>>>
>>> I've used migratepv to move oracle DBs around with no problems. I would
>>> not expect any issues with using LVM mirroring or migratepv to move the TSM
>>> DB. That is what I would do in your situation.
>>>
>>> But your comments about taking days to backup and restore your TSM DB
>>> worries me. How long does it take to backup your DB? I have a 600G
>>> allocated/415G used TSM DB. It backs up in under an hour and restore time
>>> is about the same.
>>>
>>> David Ehresman
>>> University of Louisville
>>>
>>> -----Original Message-----
>>> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On
>>> Behalf Of Roger Deschner
>>> Sent: Thursday, March 20, 2014 11:35 PM
>>> To: ADSM-L AT VM.MARIST DOT EDU
>>> Subject: [ADSM-L] Moving the TSM DB2 database (AIX)
>>>
>>> Now that TSM V5 is gone from our shop and we're all TSM V6.2, it's time to
>>> move some things around. Such as the TSM DB2 database. The manual says to
>>> do a full database backup and restore. That could take days of downtime
>>> with our 150-300GB databases, and a lot of angst, so that is not really
>>> acceptable.
>>>
>>> What I'm planning to do instead, is what I've always done on AIX. It's one
>>> of the reasons I like AIX for hosting something like TSM. That is, to
>>> basically walk the database over to the new location using AIX LVM
>>> mirroring. All this with TSM up and running, albeit with a performance
>>> impact. (It's Spring Break, so the performance impact is acceptable.) The
>>> end result will be that the database has exactly the same Unix filesystem
>>> names, path names, and file names as before, except that it will be on a
>>> nice new faster disk subsystem.
>>>
>>> Other than the obvious performance impact while AIX LVM is doing the
>>> mirroring, is there anything wrong with moving a TSM DB2 database by this
>>> method? Anybody done this and had problems?
>>>
>>> Roger Deschner University of Illinois at Chicago rogerd AT uic DOT
>>> edu
>>> ======I have not lost my mind -- it is backed up on tape somewhere.=====
|
