Time to revisit opinions on VTL's

[spiffy]

Hi all

I just want to gauge from you your opinions on TSM using a VTL.
I started a thread a long time ago and got mixed reviews, some saying TSM should not be used with a VTL others saying go for it.

We had EMC in and take a look at our environment and recommend their VTL solution, then we had IBM in and they recommended LTO4 technology...
Sepaton sent me information too

The nice thing about VTL in my opinion is potentially quick recovery time, with tape, not so much.
I actually figured out how long it would take to restore our file server (over 1TB) using our current LTO2 technology (approx 35MB/s on hardware) not allocating tape mounts, latency, using 1 drive, it will take 26 days straight to restore.
that is were VTL trumps tape, in theory it should write and read as fast as the disk, however there has to be gotchas there, bottlenecks, how much tweaking on the tsm server and or clients need to be done to keep up with fast disk.
then there is traditional tape, a tried and true solution. tapes are removable, and cheap, but slow...
i read and hear about all the goods things with both, but am having a hard time finding the bad... possibly my google skills are lax.

anywho, let me know what you all think.
Unfortunatly, our budget this year does not allocate for either solution, you win some and lose some...

 

[dangel42]

In my opinion, a VTL is not necessary. Just have TSM write directly to the disk that the VTL was/is going to use.

 

[heada]

Some people love VTLs. Others see no need for them. TSM can be configured to use them so that it is advantageous but you can configure TSM to do everything a VTL does without a VTL.

A lot of the vendors will try to sell them to you saying that it will save lots of time in your daily processing, which is not true unless you have a less-then-optimal configuration in TSM. The offer limited benefits over a diskpool for backup speeds and with collocation and active data pools, limited restore benefits. They're more expensive than tape (power, cooling and purchasing) and less dense than tape. They're also limited to your SAN/network connection for offsite vaulting whereas tape can simply be shipped anywhere.

-Aaron

 

[moon-buddy]

I would agree with Aaron and others sharing the view that there is NO need for VTL.

Vendors carrying the VTL solution sell them because they can make more money down the road even if they give it to you at half their "at cost" amount. VTLs are power hungry, requires a lot of cooling, not portable and tends to be expensive when you need to expand capacity compared to a comparable sized (physically sized) tape library. In the long run, VTL cannot accommodate too many versions of a backed up file since this would be an expensive operation.

Follow good TSM setup as pointed out by Aaron and you should have a good, and cheap, backup/DR platform.

 

[reddy]

Hi all

I just want to gauge from you your opinions on TSM using a VTL.
I started a thread a long time ago and got mixed reviews, some saying TSM should not be used with a VTL others saying go for it.

We had EMC in and take a look at our environment and recommend their VTL solution, then we had IBM in and they recommended LTO4 technology...
Sepaton sent me information too

The nice thing about VTL in my opinion is potentially quick recovery time, with tape, not so much.
I actually figured out how long it would take to restore our file server (over 1TB) using our current LTO2 technology (approx 35MB/s on hardware) not allocating tape mounts, latency, using 1 drive, it will take 26 days straight to restore.
that is were VTL trumps tape, in theory it should write and read as fast as the disk, however there has to be gotchas there, bottlenecks, how much tweaking on the tsm server and or clients need to be done to keep up with fast disk.
then there is traditional tape, a tried and true solution. tapes are removable, and cheap, but slow...
i read and hear about all the goods things with both, but am having a hard time finding the bad... possibly my google skills are lax.

anywho, let me know what you all think.
Unfortunatly, our budget this year does not allocate for either solution, you win some and lose some...

From previous experience, I would not go to VTL again. Needs additional maintenance when there is not enough ppl for support, when it comes to upgrade...vendor explains almost like a brain surgery and don't get enough support.
Usually any backups that go to VTL would eventually migrate to physical tapes for offsite storage. In your case, not sure how it helps for restore if VTL backups migrate to physical tapes. If you don't have it migrate to physical tape, your tsm data must be small or VTL size must be bigger enough.
If you need faster restore capability, think about upgrading your media generation.
There might be other causes for slower in restore...one main cause of them could be backup data scattered in different tapes if that was incremental all time etc..

 

[booman55]

I agree a VTL is not really cost effective unless it can scale. The Sepaton may be the answer if it in fact delivers what it claims. Deduplication is relatively new and the results out there about it is mixed at best.

Adding a scalable VTL device gives you the speed of disk without having a lot of the overhead of doing migration and reclamation processes. Adding a second VTL in hot site and doing replication eliminates the need for backup storage pool processes.

It all comes down to doing your due diligence in testing the right solution and of course how much are you willing to spend. But bottom line VTL's aren't just a hunk of disk anymore.

 

[Bartdo]

Vtl

Possible marketing motto: "TSM is the VTL".
So why virtualise a VTL ?
Bart