Disk or Tape: Is the debate finally over?

Which is a better solution when it comes to backing up and archiving data? Disk or tape? Had you asked this question a few years ago, people would have given you funny glares. But if you ask the question today, then be prepared for a healthy debate on the subject. The fact is that backup and archival was purely the realm of tape. There was no replacing tape in this space for a long time. Nobody even bothered to consider hard drives for the job because they were ridiculously expensive. But the very fact that disk has reached a stage where you can actually debate its usage over tape is quite commendable, and does make one wonder whether it will actually replace tape. 

It's pretty interesting to see how hard drives have steadily encroached into the tape drives' domain over a span of just a few years. Hard drive prices fell, capacities rose along with the throughput. Then vendors started offering hard drive storage as an intermediary solution to backup. So you would first back up your critical data to the disk before it finally went to tape. The logic was that since the backup window on tape was pretty long, it couldn't be used for instantly backing up large volumes of data, such as that coming from an e-commerce site. So, they put the hard drive in between the tape drives as a temporary space to store data before it was finally backed up to tape. Rest as they say was history. Disk kept becoming cheaper, encouraging vendors to offer even better disk-based solutions to back up data.

The growing popularity of disk doesn't mean that the world of tape hasn't seen any action at all. The technology being used in tape drives has been constantly improving. Today, you have tape drives that can back up data up to 800 GB compressed. The speed has also gone up with capacity, thereby reducing the backup window. Take the latest Ultrium 3 drives for instance. These can back up at speeds of up to 160 Megabytes per second in compressed mode (see our shootout this month for the results of various tape drives). 

Several debating points emerge between disk and tape from whatever we've said so far-capacity, cost, speed of backup and retrieval, technology changes, and portability. When it comes to backing up huge volumes of data, then the tape is far ahead of the disk. On the cost front also, disks just can't match. The 800 GB cartridge that we just talked of costs only around Rs 6,000, while a disk of similar capacity costs far more. However, if we take throughput, then disk is far ahead. Even though tape throughputs are increasing, they're nowhere near those of disks. The latest Serial ATA 2 spec for hard drives for instance, promises a throughput of up to 300 MBps. Likewise, disk is much faster in doing both backup and retrieval of data. Both operations can happen almost instantaneously, so you don't need a separate backup window for either. Technology changes are far more common in tape than in hard drives, largely because there are so many different tape technologies.

We've talked about some of these in our tape drive shootout this month. 

The last point on portability has mostly been in favor of the tape. It's much easier to take a cartridge and stack it in a cupboard. You couldn't imagine doing that with a disk. However, portability is now being questioned in favor of disks. The logic behind portability was that you could back up data to tape, and carry it to a safe offsite location for archival. But with the concept of disaster-recovery sites gaining ground, organizations can have exact replicas of their IT infrastructure in remote locations. These can be backed up in near real time, and the disaster recovery center can even take over should the primary center fail. This eliminates all data-management issues, such as handling the logistics of moving the tape around, and more importantly hunting for the right tape when you need to retrieve any data. It may seem like a trivial issue to retrieve data, but when you have terabytes of data backed up in hundreds of tapes, it's not an easy job finding that one tape that has the data you're looking for desperately. Likewise, the cost of logistics can also be pretty high. Data archival, which involves putting away data because of a govt. regulation or historical value was also something where tapes were preferred. But even here, vendors have started offering disk-based solutions. The case in point is EMC promoted Content Addressed Storage technology, which uses disk-based storage for archiving data.

Given all these points, the disk has come a long way and taken a fair share of the tape backup market. But there are still lots of organizations that can't afford remote disaster-recovery sites, and the higher acquisition costs of disks. They still prefer tape and are willing to have longer backup and retrieval windows. But if disk costs continue to fall, then we may be in for a surprise.

Anil Chopra, Associate Editor