Just type the words ‘tape is not dead’ into Google search and you’ll get over 17 million hits.
Plenty of industry commentators and analysts have blogged, written and postured on the subject for some time.
Even disk manufacturers have stopped telling people “tape is dead”. They have realised that they have to play nice with the tape vendors now. You might well ask why, and the answer is that compliance, regulation, power and air conditioning, and the current bogeyman, carbon footprint, have all conspired to limit the use of disk in the backup process to an interim but necessary step before archiving to tape. The bottom line is that tape has been around a long time, it is well tested and is still the best end game for data storage today.
It is still the largest capacity and best storage technology for longish-term data backup. At rest (sitting on a shelf, for instance) the data residing on a tape cartridge is not directly consuming power, although some temperature and humidity control is necessary for longer periods of storage, and properly managed tape is as reliable as, or better than, disk as a backup device.
However there are a couple of things that have transpired to re-popularise tape in the backup process, in particular when considering the dominant LTO tape technology.
LTO3 was the first tape drive that exceeded the ability of backup software, servers and networks to extract small files and write them to tape, in a continually ‘streaming’ job. Larger files are not a problem, but when tape hits a bunch of small files it can slow the back data throughput from MB/sec down to small KB/sec. This can make the whole job take hours longer than it has to and, worst of all, causes the LTO tape drive to stop and start throughout the job (sometimes called ‘back hitching’), inducing unnecessary wear and tear on the drive and, more importantly, on the tape media where the business-critical data may lie. Streaming a job that’s writing to tape from the start to the finish ensures that the tension of the media is consistent, and is the best way to ensure a long shelf life for magnetically recorded data.
Introducing a disk storage system as a first backup target in a disk-to-disk-to-tape process goes some way in fixing this problem, as disk is a random access device and can have more than one write stream (backup job) done at one time. This is why tape began to fall out of favour. Writing small files is done at disk-write speed and several backup jobs can be run consecutively. You can recover recent data more quickly from a disk backup system, build reasonably large disk systems to have older backup sets available, and use de-duplication appliances which will allow you to store more useful data in one disk system.
However, eventually the capital expenditure budget, power usage, data centre real estate space, or increasing staff requirements will reach a point where backing up and holding all your backup data on disk forever is untenable and very expensive.
These problems can be solved by using LTO3 and LTO4 tapes (LTO5 is coming) in the final stage of your backup process for long-term information storage. This is where tape works best. Used for medium- to long-term storage, it can easily be carried offsite for archiving. Tape has now consolidated its position in the backup process, albeit as the last step, and the worldwide market for tape is now not contracting at all. Tape is far from dead.