Tape Consortium Makes The Case For Tape Storage
September 12, 2011 Timothy Prickett Morgan
Online backup and real-time replication of data to the cloud are something that most IT shops and many consumers are looking at these days as the amount of data they are handling goes through the data center roof and living room ceiling alike. But the Ultrium Linear Tape Open (LTO) consortium wants IT shops to remember that ultimately they want to have archives on tape stored somewhere secure for extra redundancy and safety.
Apropos of nothing in the dull days of August, the LTO consortium pointed out two studies that would be useful for IT shops trying to justify the cost of investing in Ultrium tape drives for data archiving. The first, called In Search of Long-Term Archiving Solution, was published by The Clipper Group and it examines the long-term cost of using low-cost disk arrays with big fat 2 terabyte disk drives compared to LTO-5 tape drives with automatic loaders over a 12-year term. The model put together by The Clipper Group has disk and tape archiving growing from 3 to 86 petabytes over the dozen years.
To cut to the chase scene, if you keep all the data on such arrays, and assume a three-year product upgrade cycle, then you end up spending $67.2 million to archive the data as it grows on disk arrays. The cost per capacity goes way down with each upgrade cycle, but the amount of data is growing faster so the disk costs keep going up and up. About 88.6 percent of the cost of the disk is for the equipment and its maintenance, with another 12.6 percent being energy costs for spinning the disks and another eight-tenths of a percent being allocated for the data center real estate. Tape, by contrast, cost $4.5 million to store the data over the same term, with only a miniscule four-tenths of a point for energy costs and 6.3 percent for floor space. This tape scenario assumes that the hypothetical data center moves from LTO-5 technology (with 1.5 terabytes of uncompressed capacity per cartridge) in 2011 and upgrades through to LTO-8 tapes (with 12.8 terabytes uncompressed per cartridge through 2022. The question that The Clipper Group report asks is a relevant one: is it worth $60 million over 12 years to have fast access to large data sets.
Perhaps another question worth pondering is having a service bureau owning the tapes and loaders and pushing this data out over a wide area network link and not invest in it at all–assuming the data can be encrypted, moved, and put on tape securely, of course. It would be interesting to see what high availability replication of the data on the servers and storage arrays would cost as well. For the same money as you get with replicated disk arrays, you might be able to get actual HA coverage.
The second tape study, called A Comparative TCO Study: VTLs and Physical Tape from Enterprise Storage Group, came out in February. VTL is short for virtual tape library, of course, and this is a disk array that is tweaked to present itself as a tape drive to a system, allowing for faster archiving and retrieval than can be accomplished with tape. Of course, VTL arrays are more expensive over the short and long term than LTO-5 tape drives and cartridges.
The ESG white paper being touted by the LTO consortium has a number of different scenarios based on the idea that the company needs to back up 18 terabytes of live data each night. The VTL array with deduplication software is roughly 2.5 times as expensive in the first year, and over a five-year term, it works out to be 1.8 times as costly. If you want to get clever and have your tapes picked up and archived in a remote facility and have mirrored VTLs to get your data offsite as well, then the cost differential between tape and VTLs is a factor of 4.1 in the first year and 2.8 over five years. Having to pay for half of a T3 line and a second VTL ain’t cheap.
Mark Peters, the analyst who put together the ESG VTL-tape study, will be hosting a Webcast going over the findings in the report at 10 a.m. Pacific on September 27, which you can sign up for here.