Replace Library by huge raid

[genocide]

HI there,i have little question and look for some help.

Currently i have two win2003 server with 2 qualstar 412360 library and 2x 1tb raid.

In consideration of fact that the library needs a lot of maintenance, because every 2 weeks drives are polling or tapes are damaged or something else,ands in fact of performance reasons, i want to store data on a huge raid with jbods or better: my company wants to replace the library with a huge raid.

How can i realize it????

My thought is:

-creating a new diskpool with the same preferences like the ait2-pool.

-creating diskvolumes in the same amount and with the same capacity.

-after it i would set the old ait2-storage pool to read-only and i would move all data from ait2 tapes to the new diskpool.

-After moving all data from tapes, i would edit devices and remove the path, drives and the library.



Is this the right way, is anybody out there with experience in this special case??

Thanks for help or hints

Regards S.N.

 

[tsm23]

Look at Sepaton. they provide VTL's that are very high performance, and act just like tape.

 

[sthoppay]

What about the DR Plan? No offsite copy? Sounds Risky. But, it depends on kind of data you are backing up.

 

[genocide]

any other ideas?

 

[rcdelaw]

More customers are go to a virtual tape library Scenario. The Inherent risks lay with the disks. Disks fail, the question is when will they fail. The most expensive disk arrays, fail. TSM has some inherent ability for performing VTL like functions, but they aren't as sophisticated as you would need to manage your data.



Talk to an IBM Storage Rep about Virtual Tape Libraries (since you are using IBM software) or speak with your disk array vendor before you try this. The concept is simple, the execution and maintenance is not.



If you want to have a IBM Total storage rep call you, email me at [email protected], if I knew other vendors I would give you their info to, but I don't.





:grin:

 

[chad_small]

The problem with disk only based solutions is getting the data offsite. The only working disk storage solution I have seen that actually met the true business continuence guidelines we all should stick by was a ESS based TSM solution mirrored to another site to allow for true offsite protection. Since I have been working with IBM I have seen the "impossible scenario" of multiple disks in an array failing simultaneously multiple times. Its a nightmare and does happen. You could also use the simultaneous write to the copypool when going to the primary pool if you have a good connection to the remote site. Then TSM would be able to rebuild a "bad" volume .

 

[SAN-TSM]

I myself am sick and tired of tape. For large environments it is a huge PITA. At my previous job we were constantly struggling with maintaining enough scratch tapes in the library, dealing with defective media which would constantly get stuck in the drives and SAN connectivity problems with the drives (drives going polling, etc.). Your restores would go infinately faster than restoring from tape, not to mention the fact that you can virtually eliminate the overhead associated with reclamation and migrations.



I would love to setup a solution that was all disk. There are literally dozens of ways that you can setup offsite replication such as array based replication (like SRDF on EMC arrays or PPRC on IBM arrays), host based mirroring of volumes between sites or Veritas Remote Replication, and even through TSM setting up a disk pool on a remote array as the nextpool for the disk pool on your local arry.



In answer to the original question though, what I would do is setup the new storage pool as a primary disk pool and make it the nextpool for your tape pool. I would also make the new disk pool the destination for all of your backup copy groups. This will cause your backups start going to the new storage pool. When you are ready to start migrating data from tape to disk you simply set the migration thresholds on your tape pool to 0% and letting the data migrate from the tape pool.



The problems with a disk only solution is that the cost is going to be higher than tape, and replication between arrays at different sites can be slow (although it may not be much slower than the backup storage pool process).



Also, you should still be able to create offsite copies on tape if you need to. The process would probably go faster than a normal backup storagepool process, since you aren't going from tape to tape.