Everyone talks about flash, web applications, and cloud storage, but it's tape, COBOL, and mainframes that still run much of the corporate world. IBM gave TechRepublic insight into the state of tape.
A new version of IBM's virtual tape system has higher data compression rates than its predecessor, data-at-rest encryption, and 16-gigabit connections for Z-series mainframes.
Solo configurations of the TS7700 series, announced on Feb. 6, 2018, can reach almost 2.5 petabytes, while 8-machine clusters can approach 20 petabytes. Thanks to the virtualization, you can trick a mainframe into thinking that various combinations of flash, disk, and real tape are all one pool.
SEE: Quick glossary: Storage (Tech Pro Research)
Tape may not be sexy, but just like mainframes it is ubiquitous in banks, airlines, and similar industries. "Its demise has been rumored for decades," IBM storage vice-president Jeff Barber joked.
Barber said tape systems are one of Big Blue's most successful products in the last two years. He agrees with the industry notion of "flape"--flash-and-tape--which holds that enterprise storage based on traditional hard drives may become rare as flash becomes more affordable for data that's needed quickly, and tape remains the cheapest way to store archive bits. But on the cloud and software-as-a-service front, cheap disks will remain the norm because they're faster than tape and so easy to swap for maintenance, he said. Yet even cloud providers are using tape for cold-storage tiers.
Barber said more than half of IBM tape drive sales are of industry-standard LTO drives, which shipped in their eighth generation in the last few months. The rest are proprietary IBM drives.
LTO-8 tapes can store 30 terabytes of compressed data. Industry roadmaps indicate double that capacity in every generation up to LTO-12 at 480 terabytes. IBM announced in summer 2017 that researchers in its Zurich laboratory, using Sony tape, already invented ways to put up to 330 terabytes on a tape. That's largely theoretical, and such research typically needs a decade to reach commercialism, Barber noted.
How long that progress can continue is up to the laws of physics. Researchers in the 2000s abandoned the idea of double-sided systems because modern tape is too thin to withstand reads/writes on both sides. The extra thickness necessary to make double-sided tape reliable would be a wash because it means the overall tape would be shorter to fit inside its cartridge, negating the increased capacity.
"Sooner or later we're going to be writing on atoms and that's where it gets a little dicey," Barber observed.
Other new technologies won't take so long. For example, Z-series mainframes and tape libraries will continue using fiber connections for the next couple of years, and then IBM's emerging replacement known as zHyperLink may phase out FICON in the 2020s.
- Flash storage: A cheat sheet (TechRepublic)
- Cool storage at CES 2018 (ZDNet)
- Tape storage: The smart person's guide (TechRepublic)
- Why object storage provides a valuable alternative to tape storage for archive (TechRepublic)
- Why your company should stop neglecting data storage: 6 tips for getting organized (TechRepublic)