As Data Flourishes, So Do Agency Storage Efforts

By Jon William ToigoDeath and taxes are recognized as the two absolute certainties in our modern world. Now it might be time to add a third: the explosive growth of data, which can double every year. And as the volume of data increases, so does the demand for data storage technology. Government and industry continued to expand their appetites last year both for digital information and technologies for its storage. Purchases of data storage solutions, including server-tethered storage arrays, network-attached storage products and newcomers, and storage area networks (SANs) totaled $29.9 billion in 1999, according to International Data Corp., a market research firm in Framingham, Mass. Although the overall annual growth rate in the storage market slowed from 11.9 percent in 1998 to 8.5 percent in 1999, last year's lower rate reflected "front-loaded" financing arrangements in some prior-year purchases, according to John McArthur, director of the Storage Systems Research Group at IDC.Growth this year is expected to rebound to 14 percent, and revenue could top $34.3 billion for the data storage industry, whose leaders include Compaq Computer Corp., EMC Corp., Hewlett-Packard Co., IBM Corp. and Sun Microsystems Inc., McArthur said. And much of this growth will emanate from agencies within the U.S. government. The federal government, in fact, has been keeping pace with the private sector in data storage acquisitions, said Joel Lipkin, senior vice president for customer sales and support at government integrator Government Technology Services Inc. of Chantilly, Va."The World Wide Web, in the form of electronic government initiatives, drives a large percentage of acquisitions," he said. Lipkin also pointed to the replacement of aging storage hardware as a commonly stated reason for new storage purchases.Beginning in 1999, government-issued requests for proposals began to reflect a distinct storage technology bent, with numerous RFPs focused exclusively on storage infrastructure acquisitions, rather than system acquisitions of which storage was only a component, according to Lipkin and industry officials.In many cases, government agencies leap-frogged the RFP process altogether by using existing, "pre-competed" purchase orders or the General Services Administration schedule so that agencies could more quickly procure needed storage hardware and software.The appetite for storage is widespread, said Lipkin, who noted that "three or four years ago, everyone knew where the large storage requirement pockets were: big consumers of information, such as the intelligence community, the law enforcement agencies and the [Internal Revenue Service], for example." That contrasts with the purchasing patterns of today, in which just about every office and agency seems to have become part of the storage acquisition feeding frenzy.But the correlation between increased storage requirements and increased integrator profits is not as clear-cut as it might seem. While sales of traditional, server-attached storage solutions often require the additional assistance of integrators in configuring and deploying both server host and storage array hardware, network-attached storage (NAS) products typically are designed for plug-and-play and require a minimum of revenue-producing consulting services.What NAS products fail to produce in service-oriented revenue, however, they might offset through sheer sales volume, according to Lynn Corddry, district sales manager for the federal sector of leading NAS vendor Network Appliance Inc. of Sunnyvale, Calif. Corddry compared NAS boxes to cockroaches: "You put one in the closet, and the next time you open the door, there are three or four. NAS products have a tendency to multiply."Network Appliance has doubled its revenue in the government space over the past year, according to Corddry, who added that the company began to develop the federal market only within the last two years. Over that period, Network Appliance has boosted its direct sales force to eight, a number that will double in May. The company also has signed up seven resellers since autumn. The bolstering of the federal sales effort has been driven by trends that Network Appliance and others said have seen government storage acquisitions move away from server-tethered storage arrays and toward network-attached storage products.XXXSPLITXXX-Storage solutions proposed by resellers and integrators to federal government clients typically are constrained by the client's perceived needs as well as by the existing infrastructure at the client site, according to Dan Carson, vice president at Open Systems Solutions.While Open Systems Solutions might have preferred to recommend a storage area network, Carson said, "the absence of a Fibre Channel network infrastructure to support a SAN deployment would have made a SAN proposal much more costly."Carson's view was supported by Mark Vaccarino, director of systems integration and technology with Bitco/DSS of Ashburn, Va., a value-added distributor. Fibre Channel storage area networks offer huge benefits ? from scalability and ease in upgrading to easy maintenance ? "but they are not solutions for everyone," Vaccarino said. The cost difference between a SAN and a NAS is large. A small NAS product could cost about $1,000, while a SAN costs $50,000 to $70,000, Vaccarino said.Those cost distinctions were not lost on Charles Cosner, facility manager at NASA Goddard Space Flight Center in Greenbelt, Md. Over the past two years, he has deployed four Network Appliance NAS products, delivering more than a terabyte of storage per unit.Originally, a compelling feature of the product was its support for Goddard's existing ATM network ? an older fiber optic network was replaced because it was too slow, Cosner said ? and the ease with which Cosner's own technical team could install the NAS products and place them into use.Late in 1999, Cosner purchased two Network Appliance NAS arrays to facilitate the processing of telemetry data from the Tropical Rainfall Measuring Mission satellite, launched in 1997. Originally a three-year mission, the Tropical Rainfall Measuring Mission has proven so popular and successful "that it will continue to run until the satellite falls out of the sky," Cosner said.Telemetry data received from the satellite is reprocessed by NASA engineers using specialized algorithms to produce useful products for scientific research. The products will be stored on an AIT Tape Library from Qualstar Corp. of Canoga Park, Calif., according to Cosner, with the Network Appliance NAS devices "used as an online buffer to the near-line storage."This configuration, he added, is intended to enable fast access by 50 to 60 NASA scientists locally who write the algorithms for reprocessing the data. A large-scale disk storage buffer was required for the solution because of the size of the data sets with which the scientists work.Four gigabytes of telemetry provided by the Tropical Rainfall Measuring Mission satellite per day are used to produce daily products of 20 gigabytes or more, Cosner said. A year's worth of products can exceed seven terabytes.The Network Appliance NAS arrays also provided a means for distributing large quantities of products to the Goddard Distributed Active Archive Center. Individuals interested in ordering data for research purposes can order the data from the center via the Internet.Goddard's Network Appliance systems have been acquired under the Software Engineering Workstation Procurement program, "a contract vehicle for acquiring pre-competed goods," Cosner said. While he acknowledges the availability of other storage solutions, Cosner said NAS serves well at Goddard. "It is a highly deployable solution that we can install ourselves," he said.XXXSPLITXXX-While NAS may provide an effective vehicle for meeting "incremental storage requirements," NAS solutions "do not scale to meet growing needs," said Aileen Black, the district manager for civilian agencies at EMC Corp.'s Federal and Commercial Division in Bethesda, Md. The federal budget process requires agencies and departments to deploy storage infrastructures that enable "each new storage deployment to be a step forward rather than a step sideways," Black said. She cited Dwight Beeson, the Library of Congress' systems engineering group chief, as one of a growing cadre of "visionary information infrastructure advocates" who are endeavoring to create highly sharable storage infrastructures, better known as SANs.While Beeson accepted the characterization, he described his vision as less an epiphany than an evolution."Since 1993, we have had a compound data storage growth rate of about 120 percent per year. We were getting along well with server-tethered storage until we reached the eight to 10 terabyte range," Beeson said. "We started to have difficulty managing growth and connectivity. "We had 17 servers, and everything worked, but it was difficult to move several terabytes of data," he said. The library also encountered distance limitations on some storage aspects, Beeson said.Addressing the problems first led Beeson to the Fibre Channel protocol as a server-storage array interconnect, and later to switched Fibre Channel and SANs. Using equipment purchased from EMC under the General Services Administration schedule in July 1999, the Library of Congress obtained a raw storage capacity of 61 terabytes (45 terabytes usable after configuration) that can be shared among a number of IBM RISC 6000, Sun Microsystems and Windows NT servers. Also included was a special Media Server, a 128-port Fibre Channel SAN switch, storage management software and a Sony Petabite Digital Linear Tape library.The current platform supports the American Memory project, which involves the digitization and storage of more than 5 million articles from library collections that will be made available for public access via the Web.Additionally, the SAN platform stores legislative research data, legislative process tracking data from the Library's THOMAS service, copyright registration data, catalog information, and conventional internal applications data from human resources, payroll and other systems.The EMC "Enterprise Storage Network" SAN solution was selected, according to Beeson, in part because of its support for IBM AIX and because no vendors other than EMC were tested and certified in their support of Fibre Channel SANs at the time the decision was made. Initial implementation began in July 1999 and concluded by the end of the year.The implementation effort did experience a few bumps, perhaps owing to the relative youth of SAN technology, according to Beeson. "The fabric switch was not a straightforward implementation," he said, "and it took awhile for EMC to get their AIX drivers working." He attributed other problems encountered in the solution to the limited support that current operating systems provide for SAN storage architectures.The operating system looked at the storage area network as a small computer system interface, Beeson added, but the EMC array is segregated into many virtual hard disks. The operating system limited how many hard disks could be maintained on one port."It was touch and go for a while. Before we split out our collection across two different servers, it was taking 1 1/2 hours to boot all the drives on a server," Beeson said. Despite these issues, Beeson said he would like to expand the EMC Enterprise Storage Network in the future."My vision is to get all of Capitol Hill onto a shared SAN," he said. "Additionally, we have another effort moving forward involving the preservation of vinyl recordings and nitrate-based motion picture films [by converting them into digital information]. Early estimates of the storage requirements are in the petabyte range." A petabyte comprises 1,024 terabytes. In addition to scalable capacity and ease of management, SANs have a longer sales cycle than either server-attached or NAS storage solutions, which makes them especially appealing to integrators, said Bitco/DSS' Vaccarino.Customers will need an integrator to figure out the complex technology, how to provide access and security for data in a SAN. Education requirements for Fibre Channel SANs are also greater than other storage technologies.In the final analysis, Vaccarino said, SANs may scare away some potential customers. However, with the market for SANs growing at "a huge pace," integrators, "especially EMC, [Government Technology Services], Lockheed Martin Corp., and numerous smaller firms specializing in storage," are coming to view SANs as a potential gold mine for revenue. "SANs are mysterious technology to many," Vaccarino said. "And where there's mystery, there's [profit] margin."XXXSPLITXXX-Source: International Data Corp.