ADSM-L

Re: [ADSM-L] deleing data from a containerpool

2016-08-16 02:35:11
Subject: Re: [ADSM-L] deleing data from a containerpool
From: Stefan Folkerts <stefan.folkerts AT GMAIL DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 16 Aug 2016 08:33:09 +0200
Yes, I too have noticed this and it is something to keep in mind.
At the same time, I think almost everybody using this pool will be using
SSD's for the database the impact will be overseeable.
But the directory containerpool is still the best thing to happen to
Spectrum Protect since replication came along if you ask me. great
performance increase over fileclass restores, no more stopping reclaims
during the day to increase restore performance, no more messing with
numopenvolsallowed and reclaim values and number of  processes to optimize
daily operations and restore speed...oh, and compression that saves an easy
30-50% storage and license cost on top of the deduplication!




On Mon, Aug 15, 2016 at 11:20 AM, Loon, Eric van (ITOPT3) - KLM <
Eric-van.Loon AT klm DOT com> wrote:

> Hi all!
> After doing some extensive testing with a directory container storagepool
> I noticed a significant change compared to the old traditional storage
> pools.
> In a traditional storage pool TSM stores a file like an object. In most
> cases one file is one object as far as I could see. Deleting this data is
> very fast: a delete filespace runs rather fast because TSM only has to
> delete the objects. So deleting a large database client with multiple TBs
> takes a few seconds or maybe a few minutes.
> When you are using a container storage pool everything changes. Files are
> still stored as objects, but objects are split into chunks. The average
> size of a chuck is approx. 100 KB and TSM performs dedup on this chuck
> level. So if you now delete a large file, TSM has to inspect every chunk to
> see if it is unique or not. When it is unique it will be deleted, otherwise
> not. If you delete a file which is for instance 40 GB in size, TSM has to
> inspect around 420,000 chucks before the object can be deleted. I noticed
> that this takes several seconds to complete, so one has to take into
> consideration that the deletion of large clients requires significant more
> time to complete than one is used to.
> Deleing a client with little more than 1 TB of Oracle data was running for
> more than 20 minutes. So a delete filespace for really large database
> clients can run for hours! Has anyone else noticed this behavior too?
> Kind regards,
> Eric van Loon
> Air France/KLM Storage Engineering
> ********************************************************
> For information, services and offers, please visit our web site:
> http://www.klm.com. This e-mail and any attachment may contain
> confidential and privileged material intended for the addressee only. If
> you are not the addressee, you are notified that no part of the e-mail or
> any attachment may be disclosed, copied or distributed, and that any other
> action related to this e-mail or attachment is strictly prohibited, and may
> be unlawful. If you have received this e-mail by error, please notify the
> sender immediately by return e-mail, and delete this message.
>
> Koninklijke Luchtvaart Maatschappij NV (KLM), its subsidiaries and/or its
> employees shall not be liable for the incorrect or incomplete transmission
> of this e-mail or any attachments, nor responsible for any delay in receipt.
> Koninklijke Luchtvaart Maatschappij N.V. (also known as KLM Royal Dutch
> Airlines) is registered in Amstelveen, The Netherlands, with registered
> number 33014286
> ********************************************************
>