Just back from the FUJIFLM Global IT Executive Summit, with a side trip to Spectra Logic's Analyst Day, and I am up to my ears in fresh data. The annual Fujifilm event sported one of the best signal-to-noise ratios of just about any tech conference in recent memory. As always, I left with pads of notes from all of the informative sessions.
The big news, of course, in the Fujifilm domain is the continuing advance of tape technology and tape storage capacity. By that, I mean the constituent technologies keeping linear digital storage in the game despite the hostile rhetoric and disinformation campaigns of silicon and disk storage-makers.
So far this year, we have seen a demonstration of a 220 TB capacity on a single LTO Ultrium-style cartridge courtesy of Fujifilm and IBM. It will be a few years until products actually appear in the market using the underlying technologies for perpendicular magnetic recording (PMR) on tape media, improved servo tracking and head positioning hardware and software, and the like -- but not that long.
Barium Ferrite drives tape storage technology advances
Also, LTO vendors have adopted Fujifilm's Barium Ferrite (BaFe) data storage magnetic particle layer technology and have mapped at least the next three generations of their drives to the curve of steady improvements in track and areal recording densities that are enabled by the technology.
Tape has been enjoying a significant renaissance over the past few years, in part because of the ability of BaFe coatings to store a bit using a smaller amount of turf while still enabling its signal to be heard above the din (thermal and electromechanical) of the drive. In the next cycle of development, this same technology will leverage the unique properties of BaFe, when subjected to a static magnetic field during manufacturing, to align its crystalline structures vertically.
Vendors improve tape storage capacity
As was the case with disk, where perpendicular recording took drives from gigabytes of capacity to terabytes, when PMR is implemented on tape using BaFe, cartridge capacities will grow quickly toward that demonstrated 220 TB capacity. Until that happens, BaFe is already delivering significant tape storage capacity improvements. LTO 7 is delivering a respectable 15 TB per cartridge compressed. Other "enterprise" cartridges from IBM and Oracle that also use the media are also enjoying significant tape storage capacity improvements with each generation.
Okay, so tape storage capacity will be huge. So what? Most companies made the choice to move away from tape about the same time as they moved away from mainframes. Yet, a lot of companies attending the Summit said they regretted that decision and were in the process of reversing it. Even the big industrial farmers of the cloud world were in the room to hear the latest about tape.
So, to a growing number of users, tape improvement stats mean a lot. Tape is much less expensive to own than disk or flash and perfect for the huge quantities of data that are rarely if ever re-referenced or changed. Moreover, the Linear Tape File System is providing a painless way to grow the capacity of a NAS head "invisibly" to users by simply spanning the onboard disk and flash kit to a back-end tape library for seamless file directory listings and file retrievals from either the disk pool or the back-end tape.
For example, Dternity, an archival cloud from Fujifilm, uses a NAS gateway technology to enable both local and cloud-based tape storage for smooth corporate file archiving. At the event, they announced that the gateway appliance that once provided the bridge between disk and tape can now run as a virtual machine. So, you don't need any extra gear to get that limitless back-end storage repository you've always wanted.
Those use cases -- tape in a cloud-based archive or tape as a back end to NAS -- were not even the big topics of discussion at the event. The largest companies there, and the big cloud vendors, were keen to strategize about how they would cope with the analyst community's latest projections around data growth. Depending on the pundit you consult, 2020 will see total data generation of between 20 and 60 zettabytes (ZB). That is a number followed by 21 zeroes, or a billion terabytes per zettabyte. That's a problem.
Experts: Flash, disk storage lack capacity of tape
Where the heck do you store 20 ZB to 60 ZB of data? Experts like Tom Coughlin and Fred Moore testified that there will not be enough capacity in either flash or disk storage to store all of the bits -- that conclusion considers all of the stuff already deployed, the kit in development or on the street, and factors in the overall capacity of the industry to create more stuff. Too many bits, too little capacity for silicon or disk, period.
The alternatives are simple. Either discard a lot of data that is being created (as they currently must in Large Hadron Collider experiments, for example) or find some alternative storage mechanisms.
Despite Facebook's recent interest, optical storage still looks like a dead end for mass storage. That industry has finished healing from the HD DVD versus BluRay fratricide of the last decade and is now working to put out a BluRay disk with 300 GB "at some point in the near future," then a 500 GB optical spindle "some time later," and ultimately a disk with 1 TB of capacity "at a date yet to be determined." Assuming that they achieve those goals, an assumption one would be smart to hedge, that just means that the industry will need to produce 20 billion to 60 billion of their Frisbees before 2020: another Herculean feat, especially with a dwindling consumer market.
Tape offers necessary capacity to store zettabytes
So that leaves tape. Tape is the only technology with real storage capacity growth requiring modest investments for manufacturing adjustments. The combination of Barium Ferrite and more durable substrate layers (the plastic backing that makes it a tape) offer higher density and longer tape lengths to be spooled onto the cartridge hub.
Tape is, in fact, the only hope that mankind has for storing 20-plus zettabytes (mostly new data, by the way) by the next decade. That's why, for this holiday season, I am asking Santa for a new tape library. Call it a survival strategy for the coming Z apocalypse (that's Z as in zettabyte).
It is worth noting that Spectra Logic has diversified its product portfolio to add another augment to tape storage. Its ArcticBlue appliance leverages Seagate's latest shingled magnetic recording (SMR) drives that can provide, according to Spectra Logic spokespersons, most of the economies of tape, but with the random access of rotational media. Assuming that their drives can be pushed to a longer lifecycle via periodic power-down, they may have come up with a way to take some of the latencies out of tape-based data access. Watch this space.
The evolution of tape storage technology
Tape storage backup, recovery pros and cons
Tape storage reliability continues to improve