Knowledge Management, Mainframe Costs and Magnetic Tape

Our readers comment on past issues.

Beyond Knowledge Management
Regarding David Essex' article in the March 2002 issue, "Managing Know-how:" Managing information within an organization is very challenging, as you aptly describe. There are a few points that I believe you may have missed, however.

It's not just about having the right taxonomy/ classification technology—solutions have to be designed to fit the way people work. They need to be tailored to work across an organization. Without adequate consideration of the "human factor," no amount of the technology that you describe in your article will be successful.

GlobalWisdom's Enterprise Knowledge Management software brings together state of the art technology with an interface and workflow that facilitates and captures outputs from critical thinking processes, and produces results that make sense to people. We enable organizations to augment their existing content and knowledge management systems within an evolving and collaborative taxonomy to better facilitate critical thinking.

Keep up the great writing!

—Frederick L. Hubig
Business Development, Search and Classification Solutions

Cost of Mainframes
I noticed that your back issues (available at don't go back to 1994. In the July 1994 issue, Brian Jeffery wrote an article entitled "What Mainframes Cost." I still have a hardcopy of it. Has Enterprise Systems ever revisited this important and heated topic? Eight years is a long time, and I'd like to know if the core arguments made in the article ring true today. Manageability for distributed computing has improved greatly in eight years, so it would be interesting to see Brian (or someone else) revisit the topic. Thoughts?

—Kevin Groff
Plano, Texas

Great idea, Kevin. As you note, the metrics have changed a great deal in eight years—look for an article on this in an upcoming issue.

E-learning Exposure
I've been an Enterprise Systems subscriber/reader for a long time and it was nice to finally see an article ("Innovate or Evaporate," February 2002) about our end of the industry: training.

I share Laura Wonnacott's enthusiasm for the potential of e-learning and I'm constantly looking for best practices, attending conferences, and generally keeping my ears to the ground for anything and everything going on in this space.

—Scott McFall
VP, Business Development
Pittsburgh, Pa.

The Demise (Again!) of Magnetic Tape
Note: You can subscribe to Jon Toigo's free weekly commentary on storage, or read back issues of the newsletter, at Click on Newsletters to subscribe.

I read Jon Toigo's "Storage Strategies" newsletter column (Feb. 8 edition) on "Near-Line" storage (actually, a term I thought was trademarked by StorageTek to refer to their robotic tape drives when properly used with HSM, which was once an IBM trademark).

While I don't doubt that optical storage may have a place in the storage world, including as an archive with slower access than magnetic disk—and potentially faster than some magnetic tape approaches—I must admit I'm surprised to hear (again) of the untimely demise of magnetic tape. This is especially true because I personally know of quite a few cases of recent major purchases of so-called virtual tape systems.

These appear to solve most of the problems associated with tapes at a competitive cost—and they are marketed by some of the same companies you cite as examples of those who served as ushers at the funeral. Very strange.

I no longer remember when I first heard that tape was dead—probably about the same time I heard that companies would no longer need systems programmers. However, I expect to be retired before tape is. And unless I prove to have as many lives as it apparently has, my funeral (which I expect long after I retire) may also come first.

—Marty Stahl
Principal Systems Programmer
Kansas City, Mo.

Have you guys ever heard of StorageTek?

—Larry Thiry

First, I wholeheartedly agree that tape is not dead.

As the disaster recovery efforts in the wake of Sept. 11 demonstrated, tape remains the most versatile data recovery medium and enables, among other things, the re-hosting of data in a recovery environment to an alternative storage platform. Moreover, as I have said repeatedly, it's precisely the sharing of a tape library that prompted the preponderance of early Fibre Channel SAN deployments, and ironically, it is upon tape that companies have come to depend to cope with the vicissitudes and instabilities of SAN software and fabrics!

For the record, I back up everything to tape: An Ecrix (now Exabyte) VXA tape solution to be specific.

Regarding my failure to mention StorageTek in the piece: StorageTek was and is a pioneer in "near-line" storage concepts and certainly deserves a mention.

However, the piece wasn't about StorageTek or the several other vendors I mentioned. Rather, it was about an industry trend I see: Leveraging evolving networked storage architectures to enable the effective implementation of hierarchical storage management.

HSM is a technology that hasn't found much play in distributed computing environments up to now.

Despite obvious advantages of matching data access and update characteristics to the storage platforms used to host the data, HSM across a LAN is often considered too resource-intensive or cumbersome to deploy.

But I've always believed in HSM as a way to cut costs in storage and was excited to hear that the idea is enjoying something of a renaissance as Fibre Channel fabrics—and ultimately, IP-based storage networks—provide a high-speed infrastructure for migrating data that doesn't impact LAN bandwidth.

What types of devices could be used for near-line storage? Some vendors are introducing IDE/ATA disk arrays as lower cost alternatives to high-end platforms like EMC Symmetrix or IBM Shark or HDS Lightning arrays. Others are pressing optical disc jukeboxes to the fore.

Still others note that the increasing access speeds and transfer rates possible with high-end tape subsystems make tape an ideal medium on a cost per GB basis.

I agree with all of these solutions and see a time, very soon, when storage will be more properly allocated on the basis of policies having to do with data access characteristics and platform costs.

Apologies to StorageTek, and also to NexSAN, makers of an IDE/ATA array called ATAboy, and Network Appliance, and many others whose products I didn't specifically mention in the piece.

—Jon Toigo