Taxing Government Storage Systems

DATA Explosion

Taxing Government Storage Systems

By Ed McKenna

To keep up with consumer demand for weather information online, the National Oceanic and Atmospheric Administration's Office of Satellite Data Processing and Distribution opted to upgrade its satellite archives from older tape technology to advanced, high-capacity tape.

The office, which provides satellite data to government customers, such as the National Weather Service and the departments of Defense and Agriculture, as well as researchers and commercial firms, has seen its storage technologies pushed to the limit in recent years, said Peter Topoly, program manager for the office's Satellite Active Archives.

Beginning in 1994, the agency broadened its user base dramatically by offering the data on its Web site. Online distribution grew from 2,000 data sets in 1994 to 52,000 data sets last year. "Our target is 100,000 for this year," Topoly said. The organization also plans to add historic data, going back to 1978, to its online source.

Using new Magstar tape drives from IBM Corp. of Armonk, N.Y., the satellite archives will get a fourteenfold increase in capacity with the same data throughput speed, said Topoly. "Now we have the capability of putting 60 terabytes of data online, which is enough to put all the current and archived data online," he said.

As federal agencies expand their use of the Web and deploy intranets and client/server networks, they are facing an onslaught of data and the challenge of providing speedy, secure and accessible storage for that digital information.

Spurred by the increase of network computing, the dynamic growth of digital data has become a fundamental fact of life for the public and private sectors. Digital data capacity worldwide is projected to grow at 91 percent per year through 2001, said John McArthur, an analyst with the International Data Corp., Westborough, Mass.

"Data is exploding dramatically across all different platforms," added Scott Kozar, a program manager for the storage division at IBM. "We have seen mainframe data growth of about 50 percent per year over the past few years and upward of 80 percent per year across the open platforms," he said.

In the federal arena, "we're seeing growth rates that even exceed that," said Steve Fessler, certified storage specialist with IBM Global Government Industry, Bethesda, Md.

Fueled by this data growth, the worldwide data storage market is expected to increase from about $100 billion last year to $1 trillion by 2004, despite falling systems prices due to competition, according to the Advanced Technology Program for Data Storage Solutions administered by the National Institute of Standards and Technology, Gaithersburg, Md.

The greatest growth is occurring in the Windows NT environment. Sales of disk storage subsystems for NT are projected to grow about 33 percent each year between 1997 and 2001, compared to 10 percent per year for Unix, according to International Data Corp.

Among storage media, rigid disks were the biggest sellers, garnering 67 percent market share in 1996, followed by optical disks at 17 percent, tape at 9 percent and floppy disks at 7 percent, said Mary Bourdon, a principal analyst at Dataquest, San Jose, Calif.

Government Sales

In the government market, storage subsystems sales are boosting the bottom lines of systems integrators, resellers and vendors, including such formidable players as IBM and Unisys Corp., Blue Bell, Pa., as well as storage specialists such as EMC Corp., Hopkinton, Mass.; Storage Dimensions, Milpitas, Calif.; and Box Hill Systems Corp., New York.

Explosive data growth is forcing vendors "to find solutions in three critical areas: availability, performance and management," said Paul Battaglia, federal sales director for Storage Dimensions.

"Management of this [data is] a huge challenge both from a data integrity perspective - making sure you have backups and protection - as well as from a cost perspective," said IBM's Kozar.

"There are [also] increasingly stringent requirements for availability," he said. "A network world means people are sharing applications and getting access to applications from a broader perspective, so everybody needs access to everybody else's data." Add in the Internet and "all of a sudden there are millions of people who are getting access to your information, and you no longer have the luxury of deciding when to take your systems down by talking within your group."

Network and storage management issues can be expensive to solve, but viable solutions are emerging.

"One of the things that people are doing to deal with the high cost of management is to bring all of the data under centralized control," said McArthur. This way, organizations can make more efficient use of personnel and assets, he said, adding that a lot of organizations are choosing common enterprise storage platforms, such as EMC's Symmetrix Enterprise Storage Platforms.

Enterprise Subsystems

Unlike server-based storage, these enterprise subsystems can connect with, store and retrieve data from all major computing platforms including mainframe and open systems. EMC systems would work well in consolidated data centers, such as the Defense Information Systems Agency megacenters, said Bruce Trinor, vice president of federal sales at EMC.

The DISA megacenters, for example, include a variety of platforms: IBM and Unisys mainframes, Sun Microsystems, Hewlett-Packard and NT servers, he noted. The typical Symmetrix subsystem "will accommodate a direct connect to all of those systems for data access."

A leading storage system provider, EMC has in the past year sharpened its focus on the public sector, said Trinor. Among its government customers are the Department of Justice and the U.S. Patent and Trademark Office.

Data center consolidation "isn't the Holy Grail," cautioned Philip Black, chief executive officer of Box Hill Systems. "There's some consolidation going on where you actually get jack-of-all trades, master-of-none storage," he said. "We're also seeing different departments within departments going to their own dedicated storage servers for applications that have unique requirements."

Tom Samulewicz, director of strategic planning at Unisys' open systems program

About 5 percent of Box Hill's sales are to the federal government, said Black, "We have a good number of government [customers]," he said, noting "the U.S. Army's payroll is done on Box Hill disks, [and] the Department of Justice is putting us in all their new attorneys' offices."

The company almost exclusively sells through systems integrators, including Electronic Data Systems Corp. of Plano, Texas, GTE Government Systems of Needham, Mass., and resellers such as Chantilly, Va.-based Government Technology Services Inc.

In addition to central control, organizations are solving their storage needs by matching existing storage technologies to their needs. "Typically when you're looking at storage media selections you're looking at a trade-off of performance vs. cost," said IBM's Fessler.

At the low end of the scale is magnetic tape, which is viewed as "somewhat of an insurance policy," said Battaglia. Because tape drives are too slow to handle data that must be continuously accessed, tape is used primarily for archiving large amounts of data to take advantage of its low cost and high capacity.

The need for greater data availability, though, is driving advances in tape technology, such as the speedier Digital Linear Tape drives developed by Digital Equipment Corp., Maynard, Mass.

Paul Battaglia, federal sales director for Storage Dimensions Top Disk Storage Subsystems Suppliers Worldwide

Optical disk storage devices such as CD-ROM drives, WORM (write once read many) drives and rewritable optical disks provide better performance and reliability than tape at a higher price. Like tape, however, these drives are too slow to store data that need to be accessed often. Because of their permanent nature, WORM drives have been attractive to government agencies, said Bourdon, noting that the Internal Revenue Service is a big user. The Department of Education, for example, opted to use IBM's WORM technology for its permanent archive for its student financial aid program.

For applications requiring high data availability, however, most organizations use disk storage or, more specifically, RAID or redundant array of independent disks technology. Introduced in 1987, RAID devices comprise a cluster of disk drives that work together to deliver performance surpassing that of a single large disk. The systems are rated at different levels between 0 and 5, reflecting a balance of performance and data protection with RAID 5 being the most commonly used.

Many, if not all, government departments use RAID storage systems. "I would venture to guess that [our RAID systems] are probably being used in all of the cabinet agencies [including] the executive office of the president," said Battaglia. Storage Dimensions products are available on large contract vehicles, such as the National Institutes of Health Electronic Computer Store Contract and the General Services Administration schedule through companies including BTG Inc., Fairfax, Va.; GTSI; and FedTek, Lake Ridge, Va.

The Department of Defense's On-Site Inspection Agency is currently deploying Storage Dimensions' RAID subsystems to handle its expanding storage needs. Responsible for monitoring of arms control reduction treaties, OSIA has a worldwide network with about 1,200 users, said Bill Earles, senior systems architect for OSIA.

With data volume growing about 1 percent a day over the past few years, the organization opted for the Storage Dimensions' Dynamic Growth and Reconfiguration RAID subsystem, which will more than double the organization's storage capacity. Capable of holding 45 gigabytes of data, the new systems use ultra SCSI connections, which are twice as fast as OSIA's previous systems, which used regular SCSI (small computer system interface) technology.

As impressive as OSIA's performance improvements are, many observers predict by the end of the year they will pale in comparison to the performance of new fibre channel systems. An American National Standards Institute communications standard, fibre channel is a fast serial bus interface standard intended to replace SCSI on high-end servers.

"The standard we have today is ultra SCSI, which offers us 40 megabytes per second throughput," said Battaglia. "Fibre promises to bring us 100 megabytes, but we're not quite there yet."

But he and many other industry officials believe it's just on the horizon.

"On the drive side, the technical difficulties have been overcome, it is just a matter of what you're going to hook it up to," said John Monroe, chief analyst for rigid disk drives worldwide for Dataquest.

Many companies are betting there will be systems to connect it to this year. For example, Unisys' new PrimeStor NAS2000 RAID network attached storage system introduced early this year is completely fibre channel.

"We skipped going to SCSI [altogether]," said Tom Samulewicz, director of strategic planning at Unisys' open systems program. The NAS2000 is Windows NT and Intel-based and is simple to administer, Samulewicz said, adding that operators don't have to go off line to add capacity.

Currently able to scale to 568 gigabytes, the system will increase to 1 terabyte capacity by June, he said. Unisys is adding the product to its GSA schedule. Meanwhile, the Essex County Registry of Deeds, Mass., has already acquired a NAS2000 for use in storing and managing land-title documents.

Sun Microsystems Inc., Mountain View, Calif., also introduced in January its new full fibre channel Sun StorEdge A5000. A pioneer in fibre channel technology, Sun co-founded in 1993 the Fibre Channel Systems Initiative with Hewlett-Packard Co., Palo Alto, Calif., and IBM, and began shipping early the fibre channel product Sparcstorage in 1994.

"As the market and the end users become aware of the fibre channel technology, you're going to see a mass migration to it," Samulewicz predicted. Aside from faster throughput, the technology allows for greater connectivity allowing you to attach to 126 devices as opposed to up to 15 on SCSI, he said. In addition, cables for fibre channel can be as long as 30 meters (coaxial) or 10 kilometers (optical) allowing storage systems to be located much farther away from the data center than with SCSI connections.

Introducing its Fibre Box in late 1996, Box Hill also is seeking an early stake in the fibre channel market. "We haven't got anybody [in government] who is using fibre channel yet, but we're seeing a lot of interest from a number of places, not the least being NASA," said Black.

The IRS, which relies heavily on storage technologies to serve the public, is having difficulties dealing with how its growing database is taxing its computer network.

"There has been more information to back up, but the technology for us has pretty much matched it," said Tom Lucas, senior adviser in the government program management office at IRS. "I think the greatest strain is on the networks. There is much more data to transfer over the large geographic distances."

"We have linked many of our systems together and use the telecommunications network to provide nationwide access first to a read-only copy of our master file data," he said "We [also] link together many of our disparate service center systems, so people in one center can access data that's in other centers," he said. However, there have been some problems with time synchronization of the data with different centers having differing data versions, he added.

Current plans to overhaul IRS information systems over the next few years are likely to correct the situation. Under the IRS Prime contract, currently being bid, there will be "a lot more centralization for us - a lot more centralized control and consistency of data," said Lucas.

As the income tax filing season gets under way this year, IRS systems again will be tested with the expected surge in electronic tax filings.

About 21 million Americans are expected to file electronically, a 10 percent increase over last year's total of 19 million or one in six taxpayers. At the same time, taxpayers are not only using computers to file, but are also going online to the IRS World Wide Web site for forms, information and advice. As a result, the agency is managing a flood of electronic data.

But the IRS is not alone. Every major government department has a Web site offering such information as airline safety records, medical research results and satellite weather data. All of which is good news for the booming storage industry.

1997 Rank Supplier 1997 Revenue ($ millions) 1997 Market share(%) 1995 Revenue ($ millions) 1995 Market share(%) 1995Rank
1 IBM 3,887 15.7 4,224 20.1 1
2 Compaq 2,900 11.7 1,270 6 3
3 EMC 2,528 10.2 1,591 7.6 2
4 Sun 1,672 6.8 865 4.1 8
5 Digital 1,640 6.6 1,253 6 4
6 Hewlett-Packard 1,599 6.5 1,203 5.7 5
7 NEC 1,102 4.5 1,190 5.7 6
8 Fujitsu 814 3.3 957 4.6 7
9 SNI 483 2 488 2.3 9
10 HDS 463 1.9 147 0.7 18
Other supplier revenue 7,625 30.8 7,841 37.2
Total supplier revenue 24,713 100 21,029 100
Channel markup 952 654
Total market revenue 25,665 21,683
Source: IDC

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

What is your e-mail address?

My e-mail address is:

Do you have a password?

Forgot your password? Click here

Washington Technology Daily

Sign up for our newsletter.

Terms and Privacy Policy consent

I agree to this site's Privacy Policy.


contracts DB