Texas-Sized Mass Storage Solution

To store millions of case files on a network, the Texas Department of Human Services had to think big. The result was a DVD mass storage solution that offered nearline access and was scalable to almost 5 TB (terabytes).

Every industry has its own set of acronyms and jargon that tends to distance itself from the uninitiated. The computer industry is an easy mark, of course, but the government is not far behind. When these two worlds collide, the potential for miscommunication is huge. However, when the Texas Department of Human Services (TDHS) sent out an SOS regarding its mass storage system, the language was universally understood.

The TDHS has 18,000 employees charged with managing the state's benefit programs that include welfare, food stamps, and long-term care. The agency has millions of case files stored on its Unisys mainframe. This solution was adequate for all but the agency's 35 statisticians who compile detailed reports for the federal government. "The federal government requires all kinds of statistics. We report on how many blue-eyed, left-handed children are involved with a particular program," quips William Stewart, mid-tier manager for IT (information technology) at the TDHS. Compiling the reports is difficult enough. When statisticians don't have access to data, however, they are as productive as a car without an engine. "We had a mainframe crunch," explains Stewart. "We were saving the files on eight STK robotic towers, and requests for information were all prioritized. Sometimes statisticians would come to work on the weekends and request information while we were doing production processing. It could take as long as two hours to load a particular request for a statistician."

Choosing DVD Over Traditional RAID
Due to legislative oversight (read: lack of budget), the TDHS could not upgrade its mainframe. Instead, the agency settled for $120,000 in funding after its original allotment of $150,000 was reduced. Stewart's first move was to reconfigure the statistics package to run on a new NT server. The NT server also had 60 GB of storage to handle the files used by statisticians. "We quickly discovered we really needed 150 GB to store a year's worth of files," recalls Stewart. "Then, the statisticians told us that some of their calculations include all reports from the last five years. To store that many files, our system had to have a capacity of 500 GB to 750 GB." The request went out to potential vendors — a 500 GB mass storage solution that was scalable to 2 TB (terabytes).

Stewart received plenty of vendor-submitted solutions, but they all called for traditional external RAID (redundant array of independent disks) configurations that were backed up by DLTtape. These solutions were viable, but DLTtape jukeboxes "are not cheap." Ringdale Inc. (Austin, TX), however, proposed a mass storage solution that was based on optical media. The company's Professional Data Warehouse System has seven magazines that hold a maximum of 910 DVDs for a total capacity that approaches 5 TB. The system is configured with four read/write drives and two read-only drives. And, in Stewart's estimation, backup media was not required because of the permanence of DVD. "This mass storage solution fit within our budget, but cost was not the primary determinant. I wanted online or nearline storage, and this system certainly qualified," adds Stewart.

Large Magnetic Cache Speeds Nearline Storage
Of the 35 statisticians at the TDHS, 10 are considered heavy-duty users. All statisticians may be logged into the system simultaneously. However, it is unlikely that more than five people would be "beating on the NT server at one time" in an effort to retrieve files. The Professional Data Warehouse System was also configured with 300 GB of RAM storage. Because most of the requested data is immediately cached, users cannot tell the difference between online and nearline storage. "The maximum time to retrieve files is less than two minutes with this system," comments Stewart. "Most of the time, however, the retrieval time is significantly quicker than that." Ultimately, the statisticians produce reports that are used by the state and federal governments. The mass storage system has improved this process, as well. Adds Stewart, "Some of the reports we produce are too large to be burned on a single CD. Requests now come to the IT staff, and we burn DVDs that can hold even the largest reports we compile. The DVD is then hand-delivered to a particular statistician."

Ed Hess