Operating Systems

My May column discussed tape Network Attached Storage (NAS)—a hybrid platform that combines a tape library with some front-end disk caching plus a tape file system and a network mount to create a really high-capacity file storage solution that doesn’t consume a lot of power. Among other things, I noted that this “NAS on steroids” configuration might be a great solution for storing less frequently accessed data sets, delivering to users who require an occasional look-see at the stored data with response times approximating that of the Web.

I’ve received a lot of email in response from companies that are using or plan to use this platform for active archiving (which is code for an archive based on a file system with ease-of-access characteristics). Mainframes would be very good at hosting such an archive, especially as IBM’s Gigabit Ethernet Open Systems Adapter and HiperSockets technologies continue to improve. They both provide connectivity between the archive and the outside world of users, and the ability to corral a distributed archive that leverages technologies such as IBM’s Scale Out Network Attached Storage (SONAS).

To wrap up this discussion, I want to build a more comprehensive value case for this strategy that may help you sell it to the bean counters in your organization. They like the business value of your proposal to be expressed in a narrative that addresses three goals: cost-containment, risk reduction, and improved productivity. The good news is that tape NAS and active archiving deliver value in all three buckets.

From a cost-containment perspective, you need only reference storage capacity growth rates and their corresponding costs—which are scary. IDC estimated in 2011 that some 21 Exabytes of external storage had been deployed worldwide. At the time, they estimated a growth rate of 30 to 40 percent per year, resulting in about 46 Exabytes by 2014. Then, when they factored in the storage requirements of server virtualization in the distributed world, they upped the ante to 168 Exabytes (300 percent growth) in the same timeframe. Not to be outdone, Gartner increased the growth rate to 650 percent in virtual server environments, predicting 212 Exabytes of external storage!

Storage is a big ticket item and a large component of the IT hardware budget. Since the preponderance of data growth comes in the form of file-based data, tape NAS provides a way to bend the storage-related cost curve by using much less expensive tape. There you have it: cost-containment. And that’s before you factor in energy cost savings achieved by spinning less disk and generating less heat in your data center.

From a risk reduction standpoint, an organized, active archive achieves compliance with most regulatory data retention requirements. Plus, using tape instead of a de-duplicating disk repository keeps you out of a potential lawsuit over the question of whether de-duplication materially alters data. SEC rules require that a full, unaltered copy of data be provided to the commission. Technical discussions over the issue of whether or not de-dupe technology alters data are irrelevant; the technology has never been validated in court and the cost of doing so promises to be expensive, lengthy, and disruptive. There you have it: Archiving to tape NAS is much less risky than using de-duplicated disk repositories that some of the other three-letter vendors are suggesting.

Finally, there’s the matter of improved productivity. I think that the case for tape NAS in this context has at least two dimensions. First, tape NAS keeps all information online and available, but much improved over other kinds of archiving. When workers need their files, they can still get them without engaging IT personnel to retrieve them. Another dimension, tied to the operational efficiency of IT, is that less work is involved in archive maintenance. Tape library operations are clearly defined and media verification, validation, and migration can be automated to a large degree. Compare that to a disk repository, where “soft errors” occur at about 1,500 times the rate the disk array vendors say, causing undetected data corruption or total Redundant Array of Independent Disks (RAID) set failures with a much higher frequency than most of us want to think about. The data loss problem is another component of your risk reduction argument, but the amount of downtime accrued to resolving RAID problems is part of the operational inefficiency of disk-based archiving. Tape NAS nips that problem in the bud.

There you have it: a business case for tape NAS on the mainframe. Go sell it.