A significant portion of the backup and recovery world still uses tape for at least a portion of their backups, so this blog post will focus primarily on the challenges that tape presents to a backup and recovery system. This is the second part of a multipart series on the challenges of typical backup and recovery systems. If you haven’t done so already, make sure to check out Part 1.
Part 1 talked about how difficult it is to properly size and purchase a backup system. In fact, it is impossible to properly size a traditional backup system without creating waste, because you’re only backing up for part of the day, and the rest of the day the system goes to waste. While this is true for any on premises backup system, it is especially true of tape-based backup systems.
I am far from a tape basher. I think tape has a continued role to play in secondary storage systems and is often unfairly maligned. If a tape drive is supplied with a sufficiently fast stream of data, it is one of the most reliable ways to store data.The difficulty is the phrase “sufficiently fast.”
Well-performing tape is a unicorn
Each tape drive has a minimum speed at which it can reliably transfer data to tape, from LTO-1’s 6 MB/s to LTO-8’s 112 MB/s. When you send data at a slower rate it can’t keep the drive buffer full, causing the drive to stop, rewind and resume once the buffer refills.When the drive goes back and forth like this, it is said to be shoeshining. The drive spends more time repositioning than it does writing, which causes the tape drive to appear slower than it is. It also wears out the tape and the drive, making both less reliable.
It is possible to some extent to redesign the backup system to work around this issue (turn off extra tape drives), but it flies in the face of what many people see as common sense. It’s also an ever-moving target. You have to constantly adjust your settings based on the speed of the incoming backups, the network, and the backup software itself. God forbid you upgrade any part of the system; you would have to completely re-tweak your settings.
This is why a well-performing tape-based backup system is one of the rarest unicorns of them all. You may get it right for a few weeks or months, then you must constantly monitor things as you chase this unicorn. This is why most backup customers have switched to disk as their primary backup mechanism; it simply doesn’t have these issues.
Tape requires an offsite storage vendor
The best tape backup in the world will quickly become the worst if the server catches fire and the tapes are right there. This is why any good backup system will send a copy of the backup off-site. This, of course, requires an offsite storage vendor.
This vendor must be managed and its operations monitored. We had software that checked what our tape vaulting vendor said they had versus what we believed they had, and we constantly had discrepancies. Not managing all of this was not an option. They of course charged a fee for this service, and the management of the vendor mentioned above had a cost as well.
Tapes can be lost
There have been dozens of well-publicized cases of backup tapes that have been lost on their way to another location. Backup tapes even get lost within a data center. It is simply a fact of life when using portable media.
This is why most backup experts recommend encryption for all backup tapes to minimize the risk to the environment if a tape goes missing. Assuming that the tape is only a copy, there is no real loss to the company other than the cost of the media itself if it’s encrypted. Simply delete the encryption keys for that tape and it’s like it never existed. But of course, the encryption system has its own costs and risks as well.
Not easy to scale
Consider the challenges mentioned in this article and the last: sizing and purchasing of backup systems, managing the performance of tape drives, managing the physical movement of media and minimizing the risks associated with that movement via encryption and management of the off-site storage vendor.
Each of these challenges becomes greater as the size of the environment grows. There is more waste when backup systems grow larger, bigger systems mean more tape drives that must be tuned for performance, and the more tapes you are sending off-site, the bigger the chance that one of those tapes will get lost. So in addition to all of the problems mentioned previously in these two articles, it’s not very easy to scale a backup system as your environment grows.
Unexpected & variable costs
A backup environment is constantly purchasing more capacity; compute, network, I/O, and storage are always constantly growing. Backup software products come up for renewal and suddenly you’re hit with a bill for several hundred thousand dollars, and your choices are to pay the bill or have an unsupported backup system.
This is why another challenge is that there are many unexpected and variable costs associated with a typical backup system. Move some of that backup system into the cloud and things become even more unexpected, since large restores result in egress charges.
The backup industry response
These last two articles have focused on the challenges that a typical backup and recovery system has had over the last 20 years or so. There have been multiple responses by the backup industry to address these challenges, and they will be the subject of the next several blogs. Each advancement addressed one or more challenges, but each also chose not to address several other challenges. In fact, some advancements even introduced new challenges. The next article will be about target deduplication appliances.