ADSM-L

Re: anyone using ATA disk systems

2003-06-09 13:13:20
Subject: Re: anyone using ATA disk systems
From: Zlatko Krastev/ACIT <acit AT ATTGLOBAL DOT NET>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Mon, 9 Jun 2003 20:07:38 +0300
--> A large disk pool does not appear useful for:
1)      Small number of nodes with large to huge size backups. TDP nodes
would be a good example of this (our case).

Switch to convinient tape backup technology. The benefits:
-       small mount time delay will be well paid by increased speed of
backups/restores. ATA disk arrays are targeted to be best at
capacity/price (big ammounts of data with average throughput). LTO-1 or
LTO-2 will be considerably faster even running on few drives in parallel.
Balancing the load over several sessions/drives will ensure you better
performance.
-       reduced storage expansion expenses. Adding a new frame with
several 100s of tape cartridges is cheaper than adding two new racks of
ATA drives. Floor space also adds to the calculation. Data tends to grow
at very unexpected rates.
-       tapes do not wear out that fast. OTOH on long-run disks are more
vulnerable to failures. In 3 to 10 years (depends on vendor/model,
environment) you will have to replace *all* disk in a running array.
-       much lower power consumption, resulting less investments for power
protection, air conditioning, etc. Lower fan noise would be some kind of
side effect.

Zlatko Krastev
IT Consultant






"Fred.Bateman AT usdoj DOT gov" <Fred.Bateman
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
06.06.2003 16:38
Please respond to "ADSM: Dist Stor Manager"


        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Re: anyone using ATA disk systems


So how are people using (planning to use) large
disk pools (either random or FILE - with or
without Sanergy)?

Migration's one process per node seems to limit
the usefulness of the large disk pool to
implement disk to disk backup.

A large disk pool appears useful for:
1)      Large number of nodes with small to
        medium size backups (small, medium,
        large and huge are dependent on your
        hardware) and you migrate to tape.
2)      Disk pool is large enough to contain
        all primary backup data and you do
        no migration to tape.

A large disk pool does not appear useful for:
1)      Small number of nodes with large to
        huge size backups. TDP nodes would be
        a good example of this (our case).

Is there some work-around for this migration issue?

Fred
-----Original Message-----
From: rbs AT BU DOT EDU [mailto:rbs AT BU DOT EDU]
Sent: Friday, June 06, 2003 8:48 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: anyone using ATA disk systems
Importance: Low


>We have also been looking at using a large diskpool.
>It appears migration only uses one tape drive per node.
>So, if you use TDP to back up 500GB to disk and then
>run migration, it will only use one tape drive to
>migrate that 500GB.
>
>Is this true?

Migration has historically run as one process per node's data,
so what you are seeing seems to say that this remains true today.

  Richard Sims, BU