When considering the reliability of our data (or perhaps more likely unreliability) many of us think of the hard drives in our servers, desktops and notebooks. And thats natural, considering that a hard disk is the most complex individual piece of electromechanical engineering in the box. However, looking at this weeks headlines, perhaps that concern is misplaced.
Take for example, the news that Seagate Technology LLC will transition its high-performance server drives down from the usual 3.5-inch form factor to the 2.5-inch footprint usually found in notebook computers. (See Seagate to Use 2.5-inch Drives in Servers for more information.)
Actually due for its formal announcement next week, Seagates forthcoming 2.5-inch "Enterprise Disc Drive" is expected to hit the market in products about a year from now, a Seagate manager said. (No doubt Seagates use of the word "Disc" here instead of "Disk" will lead to some confusion, especially for search engines. The optical market has held the rights to the letter "C" for a long time).
Although small, these drives will offer high performance as well as reliability. For example, Seagates Cheetah line has an MTBF (mean time between failure) rate of 1.2 million hours. These forthcoming drives may offer similar MTBF standings. In addition, these drives will run cooler, requiring 40 percent less power than their larger siblings, the company said. High temperatures always degrade reliability over time.
Meanwhile, Western Digital Corp. made its Serial ATA enterprise drive available in the retail channel. (See WD Raptor Drives Now Available At Retail for more information.)
Obviously, theres demand for enterprise-level reliability and performance in the prosumer segment. The Raptor offers a 10K spindle speed as well as a 1.2-million-hour MTBF.
I commiserated about such reliability with Eric Herzog, vice president of marketing at storage startup Topio, the maker of the SANSafe data replication and disaster recovery solution. He held the same title at RAID controller vendor Mylex. Bucking the trend in the Valley, the company recently secured a second round of financing.
About 10 years ago, the top MTBF available from a server drive was perhaps 300K hours, Herzog reminded me. In those days, the current levels of reliability (and speed) were only data points on a product road map. Of course, server managers cant rely on a single drive, which is why SAN and servers use RAID to really protect data. And use products such as Topios to ensure recovery.
This combination of improved engineering and manufacturing for drive mechanisms and RAID has addressed much of the reliability equation—for hardware. Yet we keep conceiving reliability as a function of hardware—and particularly storage.
Perhaps another scan of this weeks headlines will provide a different list of candidates for reliabilitys weaker links.
Or simple human error? Ask the thousands of Yahoo customers who had their Web sites pulled because of a billing error. Lets hope they had a backup on disk.
David Morgenstern is a longtime reporter of the storage industry as well as a veteran of the dotcom boom in the storage-rich fields of professional content creation and digital video.