Tape: It Ain’t Dead Yet
August 29, 2016 Alex Woodie
In 2003, the inestimable Richard Pryor got in a good, last laugh when he starred in the Comedy Central documentary “I Ain’t Dead Yet, #*%$#@!!” In some ways, you could say that magnetic tape is the Richard Pryor of the storage world: Everybody seems to think tape is already dead, but it keeps delivering the goods year after year anyway.
If your midrange shop is like most, there is probably a magnetic tape drive toiling away. That’s because, despite the repeated predictions of its death, tape still plays a critical role in the back-office IT operations of tens of thousands of companies around the world.
This wasn’t supposed to happen. But like the IBM System z mainframe and IBM i servers that stubbornly refuse to die, tape has managed to stave off death and keep a firm foothold in the enterprise.
The reason for this unlikely occurrence is relatively simple. It comes down to economics. Plainly put, tape is cheap. And when your company is attempting to store petabytes or even exabytes worth of data, cost is a big factor.
Storage On The Cheap
A recent comparison of tape storage versus disk-based storage by Enterprise Strategy Group conducted on behalf of the LTO Consortium concluded that an organization using a tape-based storage system over a 10-year period would spend 85 percent less money than had they used an equivalent disk-based setup.
Granted, the study was conducted by the biggest tape trade group in the world, and comparing tape to disk is inherently an apples-to-oranges affair. Nobody is going to confuse tape’s sequential read capability with the random access capability of a hard disk. Nobody is going to power a 10-million-transaction-per-day enterprise with a database that exists solely on tape. Obviously, hard disk drives (HDDs) spinning at 15,000 RPM, along with a generous helping of Flash-based solid state drive (SSDs), will continue to be power these sorts of systems, at least until the price of RAM drops to the point where running a transactional database purely in memory is feasible.
But tape clearly still plays a major role in short-term backups, long-term storage, and archive applications. Tape is hands-down cheaper than storing data on spinning disk, whether that disk is sitting on-premise, in a partner’s data center, or in a public cloud. Tape always has been cheaper, and it appear that will be the case for a long, long time.
Earlier this year, the LTO Consortium–which is composed of IBM, HPE, and Quantum–published a mind-boggling number. In 2015, members of the consortium shipped 76,000 petabytes (PB) of compressed LTO tape capacity, an 18 percent increase over the prior year. What’s more, over the 15 year life of the LTO program, there has been 385,000 PB of compressed storage capacity, the LTO Consortium said. Those are massive numbers that indicate the market is relatively healthy.
Tape In The i
Clearly, tape isn’t dead. While the tape market may not be what it once was, it’s still strong. And among the majority of IBM i shops, tape is still the primary way to back up data.
In its 2016 State of Resilience study, Vision Solutions found tape was the primary disaster recovery (DR), method for about 68 percent of respondents, while about 45 percent report using logical replication (i.e. traditional software-based high availability) as the primary DR method. However, tape use has declined slightly over the years (it had an 80 percent market share in Vision’s 2010 State of Resilience report) as companies move to “more nimble” technologies, Vision says.
Vision’s numbers are at odds with another survey from HelpSystems. In its IBM i Marketplace Survey for 2016, HelpSystems says about 38 percent of organizations report tape as their main form of DR, while 44 percent claim high availability (HA) software as their main DR method. That tape number was actually down substantially from the previous year, when HelpSystems said 60 percent reported tape as their primary form of DR. (HelpSystems explains the big change was due to the fact that it didn’t include HA as an option in the 2015 study, so apparently respondents picked their “secondary” DR method. Judging from the Vision report, it’s very likely that the respondents to HelpSystems survey are still using tape, even if it’s not their claim it as their “primary” DR method. The company only allowed users to pick one method).
Interestingly, according to Vision’s study, nearly 45 percent of survey respondents indicate they’re using high availability software (which aligns nicely with HelpSystems’ number), while disk-based backup and cloud services accounted for only 8 percent and 7 percent of respondents’ DR strategies, respectively.
While D2D appliance and VTLs are growing, the biggest threat to tape’s long-term viability arguably comes from the public cloud services from Amazon, Microsoft, and Google. These companies buy huge amounts of storage at a deep discount compared to the average midrange shop, which gives them massive economies of scale to drive down the cost of storage for everybody on the cloud. With the cost of network bandwidth dropping, it’s becoming more and more viable to use off-site cloud storage as the main long-term repository for backups, cold, and archive data.
Tape Is Long Term
The key question for tape is whether it will maintain its big byte-per-dollar advantage over time. Insight into tape’s future can be found in Information Storage Industry Consortium‘s recently released 2015-2025 International Magnetic Tape Storage Roadmap.
In the roadmap, INSIC states: “Between 2003 and 2009, the areal density growth for HDD was about 35 percent per year. . . . More recently, between 2009 and 2015, the rate of areal density scaling has decreased to about 16 percent per year. This slow-down in areal density scaling has been partially compensated for by an increase in the number of platters and heads in an HDD, however, the current rate of HDD capacity scaling is still much lower than historical rates.
“Looking to the future,” INSIC continues, “there is considerable uncertainty regarding future scaling rates of HDD due to the challenges associated with overcoming the super-paramagnetic effect and the uncertainty over the timing and the eventual success of the introduction of new technologies needed to continue HDD scaling.”
“In contrast,” the group says, “state of the art tape drives operate at areal densities that are more than two orders of magnitude smaller than the latest HDDs. It should therefore be possible to continue scaling tape technology at historical rates for at least the next decade, before tape begins to face similar challenges related to the super-paramagnetic effect.”
The current generation LTO-7 tape technology offers a native capacity of 6 TB per cartridge. INSIC expects the uncompressed capacity of tape to increase to 16 TB by 2017, 32 TB by 2019, 63 TB by 2021, 125 TB by 2023, and 248 TB by 2025, representing a growth rate of 41 percent per year. These numbers roughly correspond with the forecast for growing LTO technology.
To be sure, there are challenges to scaling tape. The INSIC says it will be difficult for tape manufactures to continue increasing the number of data channels in the recording heads of tape drives, with an expected increase from 32 tracks per head currently to about 60 channels per head in 2025. Minimizing wear and tear on the recording heads is another area of concern.
But all in all, tape manufacturers don’t face the same types of “heroic measures” that disk-drive manufacturers face as they come up against physical limitations of magnetic storage on a slim piece of film.
“While clearly there are many challenges associated with scaling linear magnetic tape systems to operate at cartridge capacities in the range of 250 TB,” INSIC concludes,” there is also significant experimental evidence that these challenges can be overcome by means of skillful engineering.”
We got a glimpse into what that skillful engineering might look like in 2014, Sony shocked the storage world when it announced a new tape technology that offered an areal density of 148 GB per square inch, a record for tape, and 70x more dense than LTO-6. The big breakthrough was a vacuum thin film forming technology that’s able to form extremely fine crystal particles.
While IT prognosticators will continue to say that tape is dead, those claims are far from accurate. In truth, tape remains a critical part of the modern IT operation, relied upon daily by more than half of the IBM i installed base. Just as Richard Pryor continued to make us laugh after some people assumed he was dead, tape keeps chugging along, occasionally messing up, but mostly getting the job done in a reliable and economic fashion.
Tape isn’t best for all types of data, but for backup and long-term storage, it’s still king. For these workloads, you should closely analyze the real motivation behind anybody encouraging you to dump tape for other technologies.