The digital media market is currently experiencing the largest explosion in data growth the world has ever seen. Technological advancements and increased demand for HD and 3-D content have created exponential storage requirements for content creators, owners, producers and distributors. With this exponential growth comes the inevitable challenge of effectively storing large amounts of data while still maintaining production efficiency.
There has been an aggressive argument made over the past several years, albeit mainly promulgated by disk vendors, that all storage — including archival — should be managed by using disk as the primary storage medium. Despite this dubious but loudly proclaimed view, an increasing number of organizations are opting to purchase data tape to effectively achieve their storage goals. This reality was further validated by an IDC report that estimated that there are approximately 160EB of digital data in existence, with about 1.2ZB expected by the end of 2010, and as of today, about 80 percent of that digital content resides on data tape(1). So that begs the question: Which should an organization choose, disk or tape? The answer is both.
Although disk is a standard component of most backup and archiving strategies, the push for “disk-only” architectures is experiencing increasing opposition. A recent study, conducted by the Fleishman-Hillard Research for The Linear Tape Open (LTO) Program, showed that the number of companies deploying a disk-only storage strategy in their environments declined by a surprising 13 percent last year. The study also found that 58 percent of the companies still using a disk-only strategy currently have plans in place to reintroduce tape for interim storage, and an even greater number, 68 percent, plan to at least reintroduce tape for long-term archival and retention purposes(2).
The prominent tape format used today is Linear Tape Open (LTO). LTO was developed and is currently managed by a consortium of companies, including IBM, HP and Quantum, with significant input from other companies as well. The open architecture makes the adoption of LTO more appealing to organizations as compared with the proprietary tape formats offered by some vendors. The open format allows multiple vendors to manufacture both the tape media and drives, which helps drive costs down through competition. This also provides a heightened sense of security to buyers because they know that they are not dependent on just one manufacturer.
LTO is a format that users can trust — the format now has a published roadmap through LTO-8(3). So, other than the widespread adoption of the LTO format, why is tape seeing such a resurgence amidst continuing developments in disk technology?
First, tape is incredibly fast. The latest implementation, LTO-5, is rated with a native throughput of 140MB/s. This means that a robotic data tape system containing eight LTO-5 drives can read and write data at well over a rate of 1GB/s, which is significantly faster than most disk systems on the market today. For a real world example, let's say a post facility or a television station tries to pull a 20GB file from an LTO-5 tape, and assuming the other variables in that environment can support the max throughput of the drive at 140MB/s, then that user will be able to retrieve the file in about five minutes. For most environments, this five-minute retrieval time is more than adequate.
Secondly, tape is extremely reliable. Perhaps equally as important as the read/write speed, users have confidence that they'll be able to retrieve files when needed. Recent research has proved that the inherent reliability of tape has increased by more than 700 percent in the past decade(4).
Additionally, the myriad of recent advancements made to both the robotic tape library systems and tape results in an exponential increase in system reliability. Today, most mechanical parts used in tape libraries are considered off-the-shelf-components, so the bulk of the R&D expenditures made by these companies go toward software development. This makes the systems more intelligent by enabling them to proactively monitor both hardware components and individual tapes, and therefore allows users to identify a potentially problematic piece of media before it results in an actual failure.
The final and perhaps most significant reason for tape's current resurgence is its low cost. According to a study by the Clipper Group, tape can be up to six times more affordable to acquire, and up to 27 times more affordable to operate than disk storage options(5). The two may be somewhat comparable when looking solely at acquisition cost, but once operational costs are taken into consideration, the discrepancies become obvious.
In the Clipper analysis, the electrical cost associated with storing 1PB of data over five years on disk amounts to $670,009 versus $25,873 for tape. And that's just electricity. Requirements for floor space, scaling, management and technological lifespan also drive up the operational costs for disk use while the operational costs for tape use remain low.
Workflow benefits
Now that we have examined the reasons why tape is assuming a growing role in digital storage strategies, it is equally important to recognize how disk is also necessary to create an efficient digital media workflow. The right balance of tape and disk use varies depending on the specific environment. It is essential for an organization to assess the value of a given file at various points in time to determine its ideal storage home.
For example, in a TV news environment, files are most valuable to the organization within about two weeks of first airing. Because a big story can break at any time, it's imperative that a stored file can be accessed as quickly as possible, and even if it's just a couple of minutes faster to retrieve from disk, that two minutes can mean more revenue to your organization. The biggest difference in speed is due to the random access nature of disk; because it takes a certain amount of time for a tape system to receive a command, retrieve the tape, load that tape into the drive, and find the first bit of information, it would not make sense to move these files immediately to tape.
Conversely, there are some scenarios where it is valuable to store large amounts of material on disk as well. For example, disk is the necessary storage medium for projects that are being actively worked. If you have numerous editors working on many stories or projects simultaneously, the amount of disk needed in that environment will be larger than one with only a couple editors and fewer projects. An organization must classify content based on time-to-access value, as well as by volume storage costs, including operational considerations. Once these decisions have been made, a storage management application can be programmed to handle the transfer of any file from one storage tier to the next with virtually no user intervention.
As we move from terabytes and petabytes of content into a world of exabytes and zettabytes, it is clear that transitioning to a disk-only infrastructure makes less economic sense than previously imagined. Savvy, cost-efficient organizations are recognizing the benefits of tape and re-examining its modern role in the digital media workflow.
So, why tape? The answer is simple. Tape technologies have advanced significantly over the past decade, equaling or surpassing many of the advantages normally attributed to disk, while retaining the considerable cost advantages of tape. The real trick, however, is finding the right balance of both disk and tape to build an optimal digital storage infrastructure.
Peter Halpern is a media and entertainment market specialist at Spectra Logic.
Footnotes:
- The Expanding Digital Universe: A Forecast of Worldwide Information Growth Through 2010
- Fleishman-Hillard Research for the Linear Tape Open (LTO) Program
- lto.org/media4.html
- Mesabi Group, Infostor Magazine, “Sense and Sensibility About Tape and Disk,” Dec. 2008
- Clipper Analysis, 2008
