Thursday, May 24, 2012

This is my inaugural post and it seems fitting to discuss tape technology since IBM celebrated the 60th anniversary of IBM tape storage last week. IBM 60th Tape Anniversary. The role of tape storage has certainly changed over the years; some of us remember when customer master files were often stored on tape and processed monthly by batch processing applications. Tape was also a favorite for the protection of data, and elaborate software applications and processes were developed to periodically copy important information from disk to tape and restore it in case the original data was lost or damaged. Now, many new disk based backup solutions exist and tape has seen its use for backup transition to less time sensitive backup and archive data.

Archive data represents a significant challenge for many IT environments. Industry consultants and storage suppliers estimate that the amount of digital archive information will continue to grow unabated at an annual rate of 50% or more for the foreseeable future. However, Information Technology budgets are not growing anywhere near as fast;  even with impressive improvements in the $/GB cost of storage, it’s very challenging to define and execute an archive data strategy that can successfully support the security and retention policies of an enterprise, while remaining within the available operational and capital budgets.

To make life even more interesting,  archive data comes in many different flavors;  for example, some is highly structured; such as extracts of relational data base files; other data may be completely unstructured, for example, video or test data. The structured data may be highly reducible, using either compression or deduplication technology. On the other hand, video or test data is likely not reducible, due to either the nature of the data, or due to the fact it has already been reduced. Ironically, files not likely to benefit from data reduction techniques tend to be very large (think Gigabytes or Terabytes) and are also rapidly proliferating.

Tape storage becomes very interesting for these non-reducible files.  It is very cost effective and its power requirements are very low (zero unless being actively accessed). It is also very space efficient with native cartridge capacities ranging from 1.5 TB for LTO Gen 5 , to 4 TB (IBM TS1140) or 5 TB (OracleT10000C). Given the attractive economics, energy and space efficiency, why is tape not more widely used for archive?  One of the inhibitors to the widespread use of tape storage has been the unique access requirements. Tape is often accessed via applications that incorporate metadata and proprietary formats on top of the end user data when writing to tape. Unfortunately, for long term archive data, this then creates a long term dependency on the tape application for access. This dependency can have many different side effects including the cost of software maintenance and support over years or even decades.

Tape usage has also been inhibited by the time required to access the information on the tape cartridge. Unlike disk drives, before information is read, the cartridge must be moved to a tape drive, the tape loaded, and then positioned to the correct file. Even in a modern tape library, this sequence of activities can take 30 seconds to a minute or more. For active archive information, where the end user may need relatively rapid access to the information, these delays may offset the financial benefits of using tape storage.

A new storage appliance offering from Crossroads Systems called StrongBox®Crossroads StrongBox addresses these inhibitors by using the new LTO® Long Term File System ™ (LTFS) for the writing of data to tape, using the CIFS/NFS file interfaces for data access and using StrongBox intelligent file management to help mask the access delays of tape storage. By leveraging standard interfaces for both ingest and storage, Crossroad Systems effectively liberates the archive data from the dependence on proprietary tape applications. In addition, the intelligent management software included with StrongBox uses file disk buffers to initially respond to access requests, helping mask the delays associated with mounting and loading a tape cartridge.

Just how much might an IT department save by deploying StrongBox in conjunction with tape storage versus a traditional NAS storage system? That was the issue addressed in my white paper “A New Approach to Lowering the Cost of Storing File Archive Information” dated April 16th A new approach. It turns out that a 2 PB archive data requirement; for non-reducible data, that the StrongBox with an IBM LTO tape library solution could save a customer up to $6.5 million over ten years.  It’s worth noting that a large number of assumptions are required, to generate a 10 year TCO, and the caveat “your mileage will vary” is certainly applicable. On the other hand the potential savings are so large that it certainly appears that the StrongBox solution may provide significant financial relief to organizations wrestling with the challenge of managing very large amounts of active archive data.

The advent of clever technologies like StrongBox, that leverage the attractive economics of tape, while growing the available use cases may remove some of the inhibitors to tape usage, and substantially increase its usage for long term archive . The next 60 years should be interesting.