LTO Group Confident in Tape’s Future
March 15, 2011 Alex Woodie
People have been predicting the demise of tape for a long time–just about as long as they’ve been predicting the end of proprietary IBM midrange and mainframe servers. But today, tape remains a vital component of data center activities, just as companies continue to rely on the IBM i and z/OS platforms to run critical business applications. And if the folks behind the Linear Tape-Open (LTO) format have their way, tape will continue to stubbornly defend its niche for decades.
It’s been just over a year since the LTO Consortium unveiled the specifications for the fifth generation of the Linear Tape-Open (LTO) Ultrium format; the second LTO format, a dual-reel format called Accelis, has long since withered into obscurity. LTO 5 gear has been available for about six months from major vendors, including the three original founders of the LTO Consortium–IBM, HP, and Quantum (which acquired Seagate and its Certance division)–and numerous other storage product vendors.
While LTO 5 was a little smaller and a little slower than expected–1.5 TB capacity versus the advertised 1.6 TB, and 140 MB per second speed versus the advertised 180 MBps–LTO spokespeople make no apologies.
On the question of speed, the LTO Consortium decided that 140 MBps was plenty fast enough for the vast majority of customers and licensees, according to Laura Loredo of HP. “There will always be a segment of the market that wants faster transfer rates,” Loredo tells IT Jungle. “But for most of them, [140 MBps] is more than enough.”
The primary reason that the LTO Consortium didn’t push to fulfill the Gen 5 transfer rate first laid out years ago in the Ultrium Roadmap is that it didn’t want to exceed the capacity of computers and disk drives to feed information to the tape drives.
“We felt that going up 50 percent from Gen 4 to Gen 5 would be too big a gap for the current hardware,” Loredo says. “We were on the limit of going too fast for the hardware. And that’s something we don’t want to do. We don’t want to be too fast for the hard drives to feed the tape.”
When tape drives can write data considerably faster than disk drives and controllers can feed the drives, the tape drives continually stop and rewind to re-position the writing head. This phenomenon is known as “shoe shining,” and the additional wear and tear that it puts on drives, cartridges, and the half-inch magnetic tape itself can shorten their useful lives.
On the question of capacity, the LTO group feels that the 1.5 TB of storage (3 TB with 2:1 compression) offered by the Gen 5 spec is plenty for most applications. When you compare the 100 GB native capacity offered by the first LTO cartridges back in 2000, 1.5 TB is a lot of data.
The Gen 5 format also offers a new goody, called the Linear Tape File System (LTFS) that, for some people, more than makes up for the 100 GB loss of actual capacity compared to the stated roadmap goals. LTFS carves an LTO Gen 5 cartridge into two partitions, one with data and one with metadata, which basically turns an LTO Gen 5 drive into a self-described storage device–a giant USB memory stick, if you like.
The big benefit of LTFS is freedom of data movement. Customers no longer need a specialized application, such as a backup and recovery utility, to read the LTFS content of a Gen 5 cartridge. Users can just plug the drive into a workstation (since PCs are not likely to sport a Fibre Channel port) and then use a Web browser or Windows Explorer to scan and pull up the contents of the cartridge. Users swapping data across platforms no longer need to use complicated TAR commands.
LTFS only supports mainstream OSes like Windows, Linux, and Mac OS, as the LTO Program has no plans to support the IBM i OS (although with a little expert finagling LTFS could probably be used to store IFS content). One of the major uses of LTFS is in the broadcast industry, where some TV stations see LTO as a replacement to the proprietary storage systems now in widespread use. Those proprietary systems cost about $2 per minute for video and audio, whereas an LTO 5 system will cost about 4 to 5 cents per minute.
While LTO is making headway in specialized industries like media and entertainment, the LTO Program foresees continued strong demand for tape systems functioning as long-term data archives for companies in all industries. So why would any sane person bet that a 50-year-old technology (as half-inch magnetic tape is) will still be around 10, 20, or 50 years from now?
The simple answer is cost, according to Quantum’s Tom Hammond. “That’s one of our top stated goals of the technology: to make sure that tape continues to be best value within the data center, on a cost per GB basis,” Hammond says. “As tape moves more to secondary or archive roles, we need to sure that we have a strong value proposition.”
LTO 5 tape drives cost about $3,000, while LTO 5 cartridges cost from $60 to $120. The per-unit costs for LTO gear has not changed dramatically over the last 10 years. If the cost curve stays flat, and the LTO Consortium makes good on its roadmap, which sees 3.2 TB per cartridge with Gen 6, 6.4 TB with Gen 7, and 12.8 TB with Gen 8, then LTO will continue to have a big cost per GB advantage over disk for a long time.
Tape also offers the twin benefits of portability and permanence, which are important for disaster recovery (DR). While disk-based story devices such as virtual tape libraries (VTLs) offer data replication over the network, that isn’t always as great a benefit as it might seem, Hammond says.
“Even if you’re replicating to another site, if you have a disgruntled employee, a virus, or any kind of corrupt data, you’re just replicating bad data,” he says. “We’ve seen many cases where customers are just relying on disk and on-line replication, where they might lose their entire business if something goes wrong.”
Indeed, according to a recent survey, many companies that had abandoned tape drive technologies are actually re-instating tape, or planning on re-instating tape. The survey, which was done by the LTO Program’s public relations firm Fleishman Hillard, was conducted in 2009.
That doesn’t mean that the LTO Program sees itself in a battle with VTLs or the latest data de-duplication devices. “We are not trying to compete with disk,” Loredo says. “You need high speed access to get to data and keep SLAs [service level agreements] with customers. After that, when you have data that doesn’t have a stringent SLA, you can put that on tape. You need a blend of disk and tape together.”
Features like AES 256 encryption, WORM protection, backward read/write compatibility, and 30-year lifespans make LTO a popular choice for long-term archives. When the specs for LTO Gen 6 are released in late 2012 or early 2013, there may even be a fun new feature, like LTFS was for Gen 5.
LTO currently owns a 96 percent share of the midrange tape market. It has basically killed the other technologies, including AIT and Quantum’s DLT, which may not be dead but is relegated to legacy mode. If the LTO Program can execute on its roadmap and deliver useful new features, there’s no reason to think that companies won’t be installing new tape systems 10 years from now.