The role of tape in disaster recovery

Despite the growth of remote backup over the network, tape is still a key element in disaster recovery (DR) plans. Pierre Dorion, data center practice director at Long View Systems Inc., answers common tape backup questions as they relate to DR.

Despite the growth of remote backup over the network, tape is still a key element in disaster recovery (DR) plans. Pierre Dorion, data center practice director at Long View Systems Inc., answers common tape backup questions as they relate to DR. His answers are also available as an MP3.

Click here to listen to Pierre's answers.

Table of contents:

>>Comparing the cost of tape and disk backups
>>The other pros and cons associated with using tape for DR
>>Restoring data in the event of a disaster?
>>DR-related testing best practices for tape backups
>>The RTOs and RPOs of tape and disk backups?

Can you compare the cost of tape backups to the cost of disk backups?

That's a very interesting question that comes up a lot. At first sight, there are a number of tools out there that vendors have been using for quite some time, known as total cost of ownership (TCO) or return on investment (ROI) calculators. These try to establish some sort of baseline as to which media is the cheapest to use.

Of course, it is always a little bit biased. You will see tape vendors come out on top with their tools and disk vendors come on top with their tools, with the real problem being that it is not really a cut-and-dry approach. It's very difficult to tag a price on the technology itself. If you're looking purely at the media or the hardware, it can be argued that tape becomes cheaper, although the price of disk has been consistently decreasing. And now with data deduplication, we're seeing even more savings in terms of the increased capacity per gigabyte, per storage.

With that being said, it's not an easy calculation because your comparisons never start with a clean slate. In many calculations, a lot of the tools assume that you need a certain amount of storage and you will therefore start putting in values that end up telling you that it will cost so much per disk or so much per tape.

Unfortunately, the big issue here is that you typically have an existing investment, so it is never a forklift upgrade. It is difficult to say that it will be cheaper if you buy disk or tape because you need to factor in the value of your existing equipment. And a lot of the time, to increase capacity on your tape library, for example, you would add a frame and a few drives and you would have more capacity. So it's difficult to compare that cost of an upgrade, to ripping everything and replacing it by disk, which would obviously be a lot more expensive.

So it's really hard to establish and on top of that, you have to consider power consumption. Obviously, when we look at a tape, the tape doesn't consume a whole lot of power when it is sitting inside the library, it's only when it's spun up that it actually uses power. Disk arrays tend to still be a little more power hungry, although now we have spin-down disk technology. So, your mileage will vary depending on your existing investments in the technology. So I wish I had a magic bullet sort of answer for you, but there really isn't one.

What are some of the other pros and cons associated with using tape for disaster recovery?

Obviously, tape has its drawbacks when it comes to disaster recovery and the one that we can highlight the most is speed. Disk technology is obviously quicker. When you look at the number of concurrent restore streams or data streams that you have, if you're using tape for recovery, you will be limited to the number of tape devices that you have available.

So from a typical restore perspective, it's usually not an issue because you're restoring data and you're mounting a tape. But when we're talking about disaster recovery, DR will imply multiple restore streams that are required at once. So, you need to make sure that you have enough tape devices. If you're using disk technology, you can have a lot of concurrent streams. That's just one of the advantages.

We need to also talk about replication from a disaster recovery perspective. When we're talking about data replication at a disk level, when you start restoring that data from elsewhere, you have to recreate or write tapes back from a vault (you can then technically start restoring data remotely). So obviously disk has a definite advantage when it comes to that.

On the other hand, cost may become an issue. Since tape media itself is still relatively cheap, some may be tempted to think that when replicating data offsite, if you're looking at the cost of your disk array, you need to multiply the disk array by two to get an offsite copy. Obviously the cost is far greater than simply just creating a copy of your tape. So that needs to be taken into consideration, which is obviously a disadvantage of using tape.

What can you do to ensure the reliability of tape to be certain that you can restore data in the event of a disaster?

Tape technology will have its own built-in error checking while it's writing data to the media. That said, if the media is damaged, there's really not a whole lot you can do until it's typically too late. This is why best practices always dictate that you should have two copies of your backup data. Still, too many people create single copies. It's nice that you have the technology in place to do error correction or error checking, but once you send your tape offsite, it's no longer yours. Somebody else is taking care of it and if you go to recall it in the event of a disaster, if your media is damaged, you'll only find out when it's too late.

Storage conditions for tape media need to be optimal, so it's really important that your media is stored in the right conditions. But above all, make sure that you always have multiple copies. At the very least, you should have two copies of your backups; one onsite for faster recovery and one offsite for disaster recovery.

Are there any DR-related testing best practices for your tape backups?

Do not assume; that's probably the first rule. Don't assume that because your backup software told you that the backup completed successfully, that it's actually usable. So it's a good idea to test restores every now and then.

Of course there is the odd restore request that will force you test every now and then when users want data back. But you should always test your restore data on a regular basis to make sure that what is being backed up is actually backed up. You should also make sure that everything that needs to be backed up is backed up. And this is not a tape-related issue, it is really a DR best practice for any backup to not assume that everything is backed up.

Can you compare the RTOs and RPOs of tape and disk backups?

Because of the nature of tape media, it kind of imposes limitations on the recovery time objective (RTO) from a speed perspective, as we talked about earlier. Your RTO may not be there with tape media, especially if you have to recall the tape from offsite, but also remember that we're talking about multiple data streams. So if you run out of tape drives, this may affect your ability to meet your RTOs because you can not restore data as quickly as you want.

From a recovery point objective (RPO) perspective, that's also important because there are only so many backups that you can take of data. You can not be backing up on a continuous basis when you're using tape. It's usually more applicable to a point-in-time-type of copy. If you're comparing that to disk, you can actually start looking at technologies like mirroring and snapshots, which allow you to meet a much tighter RPO. So if you're trying to restore data within the past five minutes, it's a lot more difficult to achieve that level of granularity with tape because of the nature of the media itself.

Pierre Dorion is the data center practice director and a senior consultant with Long View Systems Inc. in Phoenix, Ariz., specializing in the areas of business continuity and DR planning services and corporate data protection.

Dig Deeper on Disaster recovery storage