Does your virtual tape library have the performance I need?
The purpose of a VTL is to consolidate and speed up backups. If the VTL doesn't have enough horsepower, it becomes a bottleneck rather than an aid to backing up your data. You should find out if it has enough capacity to handle your backup needs in a reasonable fashion. You also need to know what your requirements are. You can usually find out how much you need by using your network monitoring software.
Will your virtual tape library vendor's offerings grow in synch with your enterprise's needs or will you be waiting for the next upgrade while straining to make due with an outgrown VTL? Responsible vendors will have realistic plans to share with their customers for two or three generations out to help them with their planning. Of course, roadmaps aren't cast in stone, but they can give you at least a general idea of your vendor's direction.
Do you have enough capacity, and can I add capacity easily?
There are two broad approaches to building VTLs. Some, like those from Copan Systems, EMC Corp. and Sepaton Inc. are dedicated appliances combining disks, processor and software in one package. Others, like those from CipherMax Inc. and FalconStor Software install their software on a dedicated server/disk array. The appliance approach is usually easier to set up but harder to expand. The server/array strategy takes more installation and integration, but it usually easier to expand by adding disks and perhaps additional processors.
This isn't absolute. Not all appliances are hard to expand and not all server/arrays are harder to install than appliances. Further, unlike a conventional storage array, a VTL in most architectures is an intermediary rather than an end point, and you can increase effective capacity by shuffling the data off to tape more quickly. This can have a significant impact on how easy it is to add capacity to your VTL when you need it, and you should take that into consideration.
Do you have enough bandwidth?
In a typical setup, the VTL is a single point in the backup chain, i.e., the data to be backed up flows into the VTL from one or more servers and out to one or more tape drives or libraries. If you have multiple servers you are backing up to one VTL, or multiple tape drives -- whether in libraries or not -- the bandwidth in and out of your VTL can become a performance bottleneck.
Which drives, libraries and software do you emulate?
The thing that separates a virtual tape library from an array of disks is that the backup server and software see it as a tape library. This is the great simplifying principle of VTLs -- provided the VTL can emulate your particular tape system -- or, at worst, a system which is supported by your backup software. On the other end, the VTL needs to output to tape in a form the tape subsystem can understand. It's particularly useful to have the VTL output in the format and tape size you're actually using.
What management features does your virtual tape library offer?
You want your virtual tape library to be as easy to manage as possible. Ideally, your backup software should be able to handle the VTL directly. However, this depends on the capabilities of the backup software. More and more backup products are VTL-aware for popular models of VTLs.
How well does the virtual tape library integrate with your backup software?
The main purpose of using a VTL instead of a collection of disk drives is that it is supposed to integrate seamlessly with your present backup system. That means playing well with your backup software. In fact, in a well-designed VTL installation, the software doesn't even realize it is now sending data to an array of disks rather than a tape.
How well this works in practice depends on how well the VTL works with your software. VTL makers go to great lengths to enable seamless integration with the major backup software. You need to make sure the backup software can keep its catalog in synch with both the virtual tape library and the tape library.
What's the performance impact of data deduplication, compression and other features?
Increasingly, backup is becoming more than shuffling data from working storage to tape. Compression has been standard on backup devices for years, and more and more of them are offering data deduplication for an even bigger space savings.
This is great from the standpoint of resources devoted to storing backups, but it requires processing power and that power has to come from somewhere. The processing has traditionally been done either on the server or on the backup device. A virtual tape library introduces a third item into the backup chain, and a third place where deduplication and compression can be done.
If you're backing up a lot of data in a narrow window, the performance impact of these operations can be significant. At the very least you don't want them to be any slower than your existing system and you'd probably prefer it be faster.
How do you support encryption?
Encryption is becoming a key feature in backup systems. There are several different ways to handle encryption -- a standalone appliance, encryption at the tape drive, encryption in the VTL -- and you need to make sure that any VTL you select will fit with your backup architecture.
You also need to make sure you compress your data before you encrypt it. Encryption reduces redundancy and makes compression ineffective. So you're probably going to want to combine encryption and compression at the same point in the backup chain. Some VTLs offer both encryption and compression; some rely on other parts of the chain to handle both jobs. Either solution is potentially acceptable. You can do encryption and compression at any place in the chain -- as long as you do them both in the same place.
How do you handle data portability?
One of the great advantages of tape is portability. With their reliance on disks, VTLs are inherently non-portable. This is one of the reasons tape is still the final resting place of data in most systems. This becomes important in disaster recovery and it gets especially interesting in a case where the data is being deduplicated at the VTL.
Re-deduplicating the data isn't a problem on a virtual tape library. But so far, deduplication is pretty much device (or brand) specific. If you have to restore your deduped data elsewhere, you usually need to have the same type of VTL at the remote location.
About the author: Rick Cook specializes in writing about issues related to storage and storage management.
Do you have comments on this tip? Let us know.
Please let others know how useful this tip was via the rating scale below. Do you know a helpful backup tip, timesaver or workaround? Email the editors to talk about writing for SearchDataBackup.com.
This was first published in November 2008