Reason(s) to NOT migrate data from a disk storage pool to a tape pool

xyzegg

ADSM.ORG Member
Joined
Jun 27, 2011
Messages
48
Reaction score
0
Points
0
PREDATAR Control23

Hi,


Dumb question: is there a good reason to NOT migrate data from a disk storage pool to a tape storage pool? I base my view on the idea that, if you can, why not migrate your data to a safe place like tapes?

Technical things to consider:

1) There are plenty of filling and scratch volumes available, but the acquisition process is slow and is not a simple contracting process;
2) Drives/library idle (most of the time);
3) Backup server idle (most of the time);
4) No restore process in progress;
5) No need to restore data in the next hours and, most likely, in the next days;
6) Volumes from the disk storage pool that use DISK device class are stored on a storage subsystem disk with RAID 5 enabled (with spare available // no storage mirroring // old storage);

7) Backups consists of ...
7.1) unstructured file data and structured data;
7.2) vital data and non-vital data;
7.3) volatile data and non-volatile data;
7.5) TiB of data per week;

8) daily change rate (%) unknown;
9) annual growth rate (%) unknown;
10) TSM administrators are inexperienced with little knowledge;


Thanks in advance,


Zegg
 
PREDATAR Control23

As in all things, "It Depends" In the environment I maintain we no longer have tape of any sort. Backups go to an nfs mount on a deduplicating NAS device (Data Domain) and are replicated to another site along with DB backups. This is much simpler to maintain than any tape/vtape environment I have worked with. Never have a tape stuck in a drive, no drives offline, no drive sharing to manage, no tapes to purchase.

In a more traditional disk pool -> tape pool environment I think it really depends on your restore requirements, and if you can meet the recover times. When I had that sort of setup we tried to keep the last two days backups in the disk pool since that was where most of our restore requests seemed to come from. Restores from tape took orders of magnitude longer due to a shortage of tape drives and very long mount wait times.

-Rowl
 
PREDATAR Control23

The issue of migrating or not to tape from disk has always been asked as part backup administration and data management. This discussion exist whether you are using TSM or something else.

For me, here are the points of consideration which may guide to make a decision to migrate to tape or not, or use something else.

KEY FACTOR: Understand and classify your data

1. Slow moving data:

If your data is slow moving and your company primarily deals with archival information, it may be worth it to move data into cheap storage like tape. Restoring is another factor to consider. With slow moving data, your SLA should tolerate a 'not-so-quick' turn-around time.

2. Data has a mix of fast and slow movement - consider storage TIERING:

With fast moving data, you can have active fast moving data left-behind on disk while slow data migrated to tape as soon as possible. This does not mean that fast moving data should not be moved to tape at all. If you are familiar with the data access curve, there will be a time that fast moving data does not need to be accessed that often and can be moved to tape. This is generally dictated by time (in days, or weeks). One caveat, you will need a bigger disk pool for this.

3. Fast moving data:

This, I believe, is where the discussion and decision swings to the crucial side. Fast moving data like credit information, bank transactions, real time data movement, etc., needs to be available and retrievable almost instantaneously. Rowl has a great argument of using 'tape less' environments with device class = file. I, too, use this type of backup and users find the retrieval process very fast and easy to manage.

In your case, I am guessing either 1 or 2 would apply. The key is really to understand your data and what your business needs are.

A good Backup/Recovery admin talks to Business owners and from there, determines how data should be handled from a backup and recovery perspective.
 
Last edited:
Top