Storage administrators must find more intelligent ways to manage their burgeoning storage infrastructure but avoid adopting the latest technology just for technology's sake.
That's the message end users heard--and often laughed at--on Wednesday during Nth Generation’s annual Summer Symposium, held this week in Anaheim, Calif.
Rich Baldwin, president and CEO of the San Diego-based storage solution provider, opened the conference by telling attendees that although storage is getting cheaper all the time, the problem is that the more you have, the more it costs to manage it. "We have to get storage so that it manages itself," he said. "If it's a SQL database, it should be able to handle backups and replications automatically because it knows it's a SQL database."
Other speakers had more fun with the discussion topic. Chet Jacobs, a technical director at Hewlett-Packard, illustrated the importance of managing an enterprise's storage and server infrastructures through anecdotes. He said he recently talked to an IT administrator who was replacing his company's VMS server with Windows-based servers. "I asked him, ‘Why? How many times did you ever boot up your VMS machine?’ He said, ‘What's that?’ I said, ‘That's something you will become very familiar with,’ " Jacobs said.
Jacobs also noted that information lifecycle management (ILM) can help eliminate duplicate files and cut storage costs. He recalled a recent download of a 2-Gbyte file from Home Depot that showcased the features of a laser-guided saw he thought might be useful at home. "I downloaded it to an HP disk. It was replicated from Boston to Colorado Springs and then copied to tape in both places,” he said. “[HP] now has 297 copies of that 2-Gbyte file that I downloaded. That doesn't make a bit of difference to [HP CEO Mark] Hurd's bottom line."
The first step to managing storage is to know how much storage a company has--though it’s tough to arrive at even a close approximation, according to Jacobs. For example, when HP and Compaq merged a couple of years ago, HP asked Compaq's IT managers what their storage capacity was, and they answered 2.4 petabytes, he said.
"We then sent out a piece of software to check. It came back saying 7.2 petabytes, with 4 percent utilization,” Jacobs said.
Speaker Steve Sicola, vice president of advanced storage architecture at Seagate Technology, warned that companies desperate for ways to better manage their storage must avoid “false prophets” and “misguided rookies.”
False prophets are technologies that promise to solve an enterprise's storage problem but won’t necessarily. For instance, Sicola said many vendors look to move storage intelligence from the host into the SAN fabric, which might improve management but also might add complexity and cost to the storage infrastructure, resulting in a lengthy requalification of the infrastructure. "The jury's still out on that," he said.
Many businesses, too, are adopting SATA hard-drive technology for enterprise-class operations but aren’t considering whether that technology really fits the bill. "SATA is a low-end interface for desktop drives, where reliability is not key," he said. "Desktop drives are designed for 20 percent utilization, while enterprise drives get beat up 24x7. You need a lot of hardware and software to increase SATA reliability."
End users may be tempted by new technology from storage startups, yet many of these fledgling companies are “misguided rookies” because they don’t see the whole storage infrastructure picture, Sicola said. "Storage startups help customers see new possibilities, but they aren't well-rounded. They may solve scale issues but not performance issues. Or they solve performance but not scale. Or they solve capacity needs at a good price, but they don't solve reliability issues,” he said.
When evaluating new technologies, users must ask vendors and solution providers questions, demand reliability and ensure they’re getting the right storage for the right application, according to Sicola. "If someone says to use desktop drives, ask what it does for your reliability," he said. "Also, think about software reliability, not just hardware."
Customers also should request a total cost of ownership analysis that spells out the cost of installation, cost vs. performance, and the price of upgrades and service. "This will help compare against someone who is selling something really cheap," Sicola said. Other issues for IT administrators to consider are what happens when they’re required to ensure that data can be kept and accessed forever, the cost of unplanned downtime and the cost of data loss, he added.
In the end, Sicola said, it’s important to remember four key words: cost, reliability, availability and performance. He gave those words an easy-to-remember acronym: CRAP. "CRAP is good," he said.