Get Users Involved, Says Yahoo Boss

BURLINGAME, Calif. -- StoragePlus -- If IT managers want to keep storage under control, they need to get non-ITers into the loop, according to David Webster, manager for IT architecture and strategy at Yahoo. (See Yahoo Honcho to Keynote StoragePlus.)

The exec, who is responsible for Yahoo's corporate storage, explained in his keynote address here today that risk analysis, data retention policies, and storage costs need to be thrashed out with business managers at the earliest opportunity. Specifically, IT managers need to work out issues such as the lifespan of data stored on tape.

"Without that rule in place, youre doomed to what I like to think of as the end of Raiders of the Lost Ark, with this warehouse that goes on forever," he explained. "They happen to be your tapes, but you may or may not know what is on them."

This scenario has gained notoriety. Last week, Chase Card Services, a subsidiary of JP Morgan Chase, revealed that it had mistakenly thrown out computer tapes containing confidential customer data, which were buried in a landfill. (See Chase Trashes Tapes.) Even NASA and the Los Alamos National Lab have suffered similarly embarrassing storage snafus. (See NASA Goes to the Dark Side, Los Alamos Fallout Continues, A Tale of Lost Tapes, and The Year in Insecurity.)

A system that indicates exactly who is using which tape resources can also prove extremely useful in helping CIOs rein in costs. (See Chargeback Plods On and NTP Adds Billing .) "Being able to expose those costs is really important," explained Webster. "Once you start charging back in this space, the world becomes a whole lot easier."Overall, Yahoo has 3 Pbytes of storage in its corporate infrastructure, which is separate from the production environment supporting Yahoo.com. (See Yahoo Opens Bandwidth Bottlenecks.) NetApp, which provides 18,000 spinning disks for corporate storage at Yahoo, forms the core of this infrastructure, although Yahoo also relies on VTLs from Sepaton, tape libraries from Spectra Logic, and networking gear from Cisco -- among other products and services.

In creating its storage network, the Internet firm has had to overcome technology hurdles, such as ensuring the speed of data backup onto tape matches what's coming out of the servers. "This is where we absolutely leapt on the VTL bandwagon," Webster said. "Getting that data onto a high-speed streaming media first, before you stream it to tape, is really important."

Yahoo's tape backups, he explains, are now dwindling. Incremental backups, if they are not going offsite, only go on disk, for example. "We have started to move into the tapeless world," he said, even though he doesn't claim that all tape will disappear from his network anytime soon.

The exec also urged IT managers to get their houses in order about ILM. (See EMC Earnings Credit ILM Uptick.) "We're remiss, for the most part, in the industry, on spending time and getting formal policies written down," he warned. Unless IT seeks feedback from its constituents early on, it can take on unwanted burdens, he noted.

The task of getting end users involved isn't easy, apparently. Intel and a host of other companies recently highlighted the difficulty of getting end-user buy-in for the data classification needed to exploit ILM. (See Intel Faces ILM Challenge, PMC Reports Q1, and Users Self-Destruct on Governance.)— James Rogers, Senior Editor, Byte and Switch