To help solve the arcane mysteries of molecular structures and human health, scientists at the University of Georgia in Athens use some of the world's most advanced technologies - including tape drives.
Each day, chemistry professor James de Haseth collects between 10MB and 100MB of measurement and observational data that will help researchers better understand mammalian immune defense, viral replication and cell growth systems. "This data is very expensive to obtain" says de Haseth. "Although it can be collected in a matter of minutes, the isolation and purification of the data can cost US$10,000 or more." His lab uses a mix of Windows 98/NT/2000, Linux, Solaris and OS/2 desktops and servers.
Protecting that data against loss is crucial to de Haseth's research, and that means backup. A few years ago, he installed multiple VXA-1 tape drives from Ecrix Corp., to replace the lab's overburdened Digital Data Storage-2 tape drives. Later, he almost lost everything. "Things went awry," he says. "I was attempting to add new features to a backup server that controlled the existing tape drives, and I made some grave errors." Fortunately, the VXA drives helped him recover much of the data.
The experience that de Haseth had underscores the need for regular, reliable data backup, and it shows how skyrocketing storage demands can overwhelm existing backup systems. This isn't just a problem for scientists; e-commerce sites generate staggering amounts of data that must be backed up. Magnetic tape is the traditional data backup medium; it's cheap but relatively slow, and per-tape capacity has historically been limited. Yet tape technology is keeping pace, as vendors continue to develop higher-performance, higher-capacity tape systems.
Three New Tape Types
Just coming into the market now are three tape formats - Super Digital Linear Tape (SDLT), Advanced Intelligent Tape (AIT-3) and Linear Tape Open (LTO). These offer Texas-size appetites for data archiving, a market that IDC in Framingham, Mass., predicts will grow 25 percent per year, reaching US$5 billion by 2003.
SDLT, a successor to Quantum's popular DLT 8000 format, stores up to 110GB per tape cartridge at a transfer rate of 10M bps, about doubling the capacity and speed of DLT 8000. Quantum projects that SDLT capacities will ultimately increase to 1TB per cartridge, while transfer rates will climb to 100M bps. And new SDLT drives will read cartridges recorded in the older DLT 4000, 7000 and 8000 formats.
Sony Electronics' AIT-3 cartridges, available this quarter, will offer a 100GB capacity per cartridge, with a transfer rate of 11MBps. AIT-3 is fully read-and-write backward-compatible with prior generations. AIT-4, expected in early 2004, will reportedly double AIT-3's capacity and transfer rate.
LTO, a competitive technology sponsored by Hewlett-Packard, Seagate Technology and IBM, will store up to 100GB per cartridge at transfer speeds up to 15M bps. Looking ahead to 2003, advances in LTO technology might see drives storing up to 1.6TB at the incredible (at least for tape) transfer speed of 320M bps. To accelerate file restoration and cataloging, both LTO and AIT cartridges include onboard memory that gives fast access to the cartridge's file index.
Several factors have triggered the growth in storage needs, most notably the Internet. The meteoric growth of multimedia enriched Web pages, e-commerce transactions and the Web's vast and ever-changing storehouse of documents account for a large part of the anticipated 500 percent growth in enterprise storage needs over the next three years.
Other causes for the increase are more basic. "No one wants to delete files any more, and files sizes are increasing significantly," says Derek Gamradt, chief technology officer at StorNet Inc., a data storage services firm in Englewood, Colo. Gamradt says he sees a need for storage that will accommodate the use of ever more sophisticated documents, including presentations, graphics and eventually audio and video attachments. The increase is also driven by the escalating demand by new businesses, such as telecommunications.
Ed Presutti, manager of network security and engineering at US Unwired Inc., a telecommunications services firm in Lake Charles, La., says his company's data storage requirements have more than tripled in the past two years. The data is generated by 1,000 internal employees and the company's 100,000-plus subscribers.
"We are now over the 1TB mark. When we link our accounting and financial systems with data warehousing and point-of-sale information, we will expand to a 3TB solution," he says.
Heading Presutti's wish list for storage improvements is faster throughput. "No matter what technology you use to back up your data, it just does not seem fast enough," he says. "We are also hoping to see tighter integration between the backup software, tape devices and storage-area networks."
To help ensure that tight integration, Presutti often asks aspiring vendors for an on-site product demonstration. "In our environment, what [a product] does in a lab and what it does on our network may be totally different," he says.
Tape's traditional popularity is partly due to its scalability. The technology lends itself to building multidrive, fault-resilient storage-area networks (SAN). Where throughput is important, as in retrieval operations, SANs can leverage the low latency of newer tape drives and offer fast access to multiterabyte data stores.
Despite all the advances and intelligence built into SDLT, LTO and AIT, they still depend on a linear data stream, so it's time-consuming and inconvenient to restore selected, noncontiguous files. To some degree, a hierarchical storage management (HSM) system can compensate. In this three-tier architecture, software automatically moves files between the various media - RAID, DVD jukeboxes and tape libraries. File transfers can be based on content, type, date or frequency of use. With HSM, new and frequently needed files are stored on RAID systems. Less frequently used files are moved onto near-online DVD jukeboxes. Rarely used files are archived to tape libraries, often at remote locations.
In what could be good news for some network administrators and a clarion of doom for data center managers, the benefits of outsourcing may encourage companies to outsource their backup operations.
William Hurley, an analyst at The Yankee Group in Boston, predicts that lower-cost, widely available broadband will increase the adoption of off-site storage and related document management services. The off-site approach could eliminate the need to assign scarce resources to on-site storage systems. So-called e-storage sites, run by firms such as Zantaz.com Inc. in Pleasanton, Calif., GiantLoop Network Inc. in Waltham, Mass., and Archive Inc. in Culver City, Calif., promise 99.5 percent uptime, two-hour problem resolution and infinite scalability.
"Outsourcing backup operations to [sites like Boston-based Iron Mountain Inc.'s] storage centers is cheaper, faster and provides more consistent results than archiving data in-house," says Hurley. "And you have outside people assisting you with the process ... so you don't have to assign even junior-level technicians to this task."
In addition to eliminating infrastructure, maintenance and re-engineering expenses, e-storage offers real-time searches and retrieval as well as on-demand replays of key transactions.
Jim McDermott, CEO of Archive, predicts that document management services offered by e-storage application service providers will enable once-static data backups not only to protect the data but also to help form a knowledge base. "Much of the data companies now use, such as purchase orders and correspondence, originates with external sources," McDermott says. "This is a way to not only store that information but to get fast, convenient access to it from anywhere."
Today, you can choose from multiple hardware and media technologies to protect your company's data. But backup technology is about to take some giant steps forward in capacity and speed, so the equipment that you commit to now may be outdated in a year.
What factors should you weigh before buying? Not the cost, says Presutti. "Price is an important factor, but so is quality," he says. "Look for a strong product from a strong company."
Millman is a freelance writer in Croton, N.Y. Contact him at hmillman@ attglobal.net.