Tape and disk service very different needs so why the battle cries against tape in backup/recovery?
Some analysts and vendors have, for years, been droning on that tape is not suited to backup/recovery solutions. The way this is typically done is with a ‘tick in the box’ side-by-side comparison. To anyone in the industry, this is a kind of “no-duh” position – disk and solid state technologies have progressed significantly in recent years and with their higher performance characteristics make them the more sensible choice for most backup/recovery needs. Oftentimes, however, this messaging is framed in a way as to question the very viability and future of tape (this is achieved by comparing disk and tape on their performance characteristics for backup/recovery). The real battleground is not in backup/recovery (disk is winning out), it is not in the archive (tape is winning out), it is in disaster recovery. So why are we not hearing from disk vendors on this front? Why the focus on backup/recovery? I maintain that tape in backup/recovery is an easy target; low hanging fruit for those with an agenda to see tape removed from the data center even if it is the appropriate technology for the job at hand.
One such provider following this trend is EMC who recently sponsored a paper by IDC, in which were detailed five reasons that firms are moving away from tape in backup/recovery. The top two reasons cited were poor backup and recovery performance. Other reasons cited were high operation and administration costs and security concerns. All they needed was one simple argument – the real reason that tape is no longer the storage medium of choice for backup/recovery is that servers seem to be multiplying faster than rabbits while often having the performance characteristics of a tortoise and tape is a sequential access, high speed medium while random access is what is needed to backup data efficiently. Moreover, restore plays into the mix too and competing technologies such as disk and solid state are outperforming tape in terms of transfer (particularly read) rates. Recognizing this makes the concept behind the paper a non-starter – everyone knows tape is largely leaving the backup/recovery sector – what benefit (other than creating FUD) does emphasizing this message bring? This paper and others like it read like a character assassination of someone who they would have us believe is on their deathbed. Whether on their deathbed or not, it’s in somewhat poor taste and certainly doesn’t provide any real useful information for customers or the market.
Luckily for the storage industry and consumers, tape is not in fact in its final death-throws. Far from it – tape is set to make something of a comeback thanks to the recent boom in data archiving. And the archive isn’t the only place where tape remains viable; tape’s biggest asset is its cost, or rather lack of it. Tape requires no power or raised floor and offers up to 10 TB of offline storage for around a hundred fifty bucks – keep it on a shelf for ten years and you have a very inexpensive way of keeping offline data around.
Aside from offering low cost offline storage, tapes other asset is portability. What does portability buy you? Well, for one, portability offers a significant bandwidth opportunity out of the data center. If a data center ejects 50 tapes per day (not uncommon) and moves all of those tapes to a location offsite, that is equivalent to transferring 150 TB of data per day offsite (assuming LTO5 with 2:1 compression). This kind of low-cost bandwidth is of real significance for disaster recovery users who need to move vast quantities of data far from its source as quickly as possible. This kind of bandwidth solution will, I think bring us full circle and play an increasingly important role in the archive too; allowing users create vast archives in a hurry and move archives between providers.
So where does that leave us, well, the simple fact is that tape and disk service very different storage requirements. A top-down comparison is useful as it shows how each can service differing needs, however a side-by-side “tick in the box” comparison of disk and tape is disingenuous, does not serve the consumer and frankly tastes a little sour.
