Disk Might Die Before Tape - Tape is getting healthier and smarter
Disk might die before tape, due to capacity and intelligence measures that could extend the life of tape storage.
Tape storage leaders say their favorite technology isn't dead and is actually getting healthier.
A recent wave of engineering may give the stalwart technology many more years of real-world usefulness.
It's not just straightforward capacity increases. High-end drives are commercially available from at 8.5 TB and 10 TB, while IBM and Sony each recently announced laboratory experiments in the mind-boggling 200 TB range.
The latest advancements focus on making tape smarter. Most notable in this trend is the Linear Tape File System (LTFS), demonstrated by IBM in prototype form seven years ago. LTFS is now becoming a mainstream technology supported in the industry-standard Linear Tape-Open (LTO) format and in proprietary enterprise systems; it allows tape data to be read as just another drive on a network. Call it the T: drive if you like; its point is to make the media irrelevant to the application using it.
IBM said the industry's LTO organization is also excited about
new intelligence in the LTO-7 format, while LTO-8, -9, and -10 were all previously roadmappped for more capacity and speed. What's different about LTO-7 is a new feature called Extended Copy. It's borrowed from the SCSI standard and lets administrators save time by copying information directly between drives without involving the host.
It's equally important to ensure the quality of the data itself.
Oracle cited several recent efforts for that cause.
In the past, If something went wrong, is it the drive or the media? There weren't many tools in the industry to do that. Companies had to schedule regular visits from service technicians to check that tapes were still good. Now,
the drive knows a lot about that piece of media on every mount, when it does a
read or a write. Before, the tape drives knew all that, they just weren't telling anybody.
Modern libraries can notify administrations about the results of tape analytic tests. For example, software can test the readability of any tape that hasn't been touched by its robot in 180 days, put aside cartridges that fail the test, and send a report to the library's humans or to another layer of storage management software.
The goal is to try to get things fixed before they actually break.
Tape libraries now perform cyclical redundancy checks. Oracle
sees the cloud as a challenge left to conquer, with an archive-centric cloud being farther down the chain of tiered storage so that even tape has a backup. But mostly, for tape intelligence, the industry is nearing the end of a successful cycle. In a twist of irony, such measures of capacity and intelligence could extend the life of the tape storage industry beyond that of disk.
The information storage consortium observed that given the recent trend of flash memory arrays nibbling at disk array installations, the increased challenges of making hard disks bigger, cheaper, or faster, and tape's surprisingly significant room for capacity growth, it could be disk — not tape — that eventually dies first.
Data centers of the future could have solid-state storage up front and reeled magnetic tape behind, all wrapped in a cloud. The amount of traditional disk that is being bought is dropping.
Physically smaller libraries, shorter times between LTO generations, easier-to-use administrative software, and upgradeable library network interfaces
are on the wish list. Disk arrays require a substantial amount of cooling,
energy, and rack space. The best thing about tape, is you can stick it on a
shelf.