This article can also be found in the Premium Editorial Download "Storage magazine: .NET server storage: Friendly or not?."
Download it now to read this article plus other related content.
NIGHTMARES 4 AND 5: unprepared, late and public nudity
Nightmares with these common themes often underscore the dreamer's performance anxieties or lack of confidence. Storage managers often report crises of confidence with respect to their acquisitions or the performance of their installations. To stop those concerns, users need better management tools, says Ken Barth, CEO of Tek-Tools, Dallas, TX.
Barth says storage resource management (SRM) tools must "cut through proprietary technology to provide useful management information," adding that "customers want to manage storage from a single pane of glass - just one management console. Once you understand the trends in your capacity utilization, and you can drill for specific information, you won't have the anxiety anymore about the decisions you are making."
But how many SRM tools can operate across a diverse SAN environment?
Mark Bregman, executive vice president for product operations with Veritas Software says the absence of effective software tools for managing increasingly complex and heterogeneous storage environments has resulted in storage that's "underutilized and over-mirrored." Until customers "get the industry to cooperate in a common management model," he says, the confidence level of storage managers isn't likely to get much better.
NIGHTMARES 6 AND 7: stuck in slow motion, without a voice and a helpless/ abandoned/crying
Shrinks say that these nightmares reflect the dreamer's inability to verbalize true feelings. As storage technology becomes more complex and as products from disparate vendors are required to work together, storage managers are becoming frustrated when attempting to get a vendor to solve their problem or even pick up the phone. On the other side of the fence, vendors are increasingly pointing the finger at other vendors saying, "It's not my problem - it's their problem." No wonder users get the distinct feeling that they're crying for help and no one is listening.
Bill Basinas, software engineering director for Avamar Technologies, Irvine, CA, and a former software engineer for Legato Systems, explains a vendor's perspective. He says, "A few years ago, Fibre Channel SANs created a lot of pain for companies. Networks were breaking down for no reason. Simple tasks like backup and restore were complicated by a complex infrastructure and a lack of knowledge about SANs [and Fibre Channel.] At Legato, we found ourselves spending 80% of the time helping customers sort out problems with their environment, rather than with our software."
According to Basinas, there was a dot-com that "had lost about a half a terabyte of user data and was attempting to recover it with Legato software. Every time they tried, they received corrupted data. Of course, they blamed the backup software."
"What we found," Basinas says, "was that they had created a file system that was too big for the Microsoft operating system. The file system was so big that just doing a file system check on server boot up would cost the company a day of server uptime." As a result, the customer had never checked the file system, "which was entirely corrupt, except for the last 20GB or so." The backup software was doing its job: The bad data being recorded on tape was, in fact, a faithful duplicate of the bad data on the disk."
Luke Fica - an Avamar software engineering manager also formerly from Legato - agreed that the complexity of the storage infrastructure was making it less possible for customers to set blame effectively. And, he adds, along with the increased complexity there will always be stupid user errors (SUEs).
Fica tells the story about "an intelligence agency within the DOD told us that all of their backups were failing and we needed to fix the problem quickly. Because of the nature of the customer's work, we needed to get our guy a top secret security clearance before he could begin his work. Once that was done, soldiers with guns escorted our guy around for two weeks while he tried to find the source of the problem. When he checked the tape autoloader, he discovered that all of the tapes in the subsystem were cleaning tapes [rather than magnetic tapes]. Apparently, a box of cleaning cartridges had been delivered and the customer had slapped a top secret label on each one [obscuring the manufacturer's label that clearly indicated that the tapes were not for recording]. It cost them two weeks of downtime to discover a simple user error," Fica says.
This was first published in August 2002