As part of the Administrative Office of Courts for the State of Alabama, the Alabama Judicial Datacenter (AJDC) is always looking for ways to be more efficient and cost-effective. We’re constantly evaluating new products, options, and ideas to see if we can benefit from them. Because we’re a state government agency, money constraints force us to do more with less. Anything we can do to reduce costs is always wellreceived and we have a long track record of getting good results when we do spend money. In almost every case, a hardware upgrade has resulted in reduced maintenance costs while boosting performance and/or capacity.
We first saw this trend developing as we migrated from a 9221 processor to a 9672 and then a z800. Each step gave us more capacity while reducing overall costs. We saw the same trend as we upgraded our DASD from multiple RAMAC Virtual Arrays (RVAs) to a single Shark. It was with this trend in mind that we started exploring the z9 Business Class (BC) processor in the summer of 2006.
The AJDC handles most of the applications and data that the Alabama court system uses for management and scheduling. We average 1 million CICS transactions each week day and several thousand batch jobs daily.
These applications handle everything from case management to child support payments, and are considered critical to the daily workings of the court system. With several thousand users needing system access, we must ensure anything we do to the system doesn’t negatively impact availability or performance. If we have an outage, the phones start ringing, and judges aren’t interested in technical details; they just want their data when they want it. Any changes we make to hardware or soft- ware must not have any negative impact on our users.
To handle the CICS loads, we run several transaction server regions consisting of a single terminal-owning region, two application-owning regions, three file-owning regions, a single Interactive Computing and Control Facility (ICCF) region, and a single region that handles requests from our Open Database Connectivity (ODBC) software and the VSE Connectors. We have several dynamic partitions that handle different batch requests with varying levels of priority. Printing a payment receipt while someone is standing at a payment window will receive a much higher priority than a standard report or form. In addition to the CICS and batch applications, we also have several Transmission Control Protocol (TCP)-related partitions running, including two TCP stacks, eight TN3270 servers, one File Transfer Protocol (FTP) server, and VSE2PDF. We also route thousands of print jobs to individual printers as they’re requested from batch and online jobs. Maintaining or exceeding our existing level of performance and availability must be a major part of any prospective upgrade.
On the hardware side of any potential upgrade, we must ensure that any upgrades will work well with our existing hardware, which includes a Fiber Connectivity (FICON)-attached Shark, two Enterprise Systems Connection (ESCON)-attached laser printers, two ESCON-attached 3490 tape drives, four FICON-attached 3590 tape drives, a HYDRA controller attached through an ESCON converter, an ESCON-attached Cisco 7206, and an ESCON-attached 2074. Our Open Systems Adapters (OSAs) are attached to a Cisco core switch over fiber connections. While we can periodically make a pitch for a new piece of equipment, we’d probably never be able to make wholesale changes to several pieces of equipment just to facilitate the upgrade of a single piece of equipment.
We considered all these aspects of our data center, user impact, environmental stability and hardware compatibility before we contacted our hardware business partner to see if the z9 processor would fit our needs and meet the criteria to justify an upgrade. After getting ballpark figures from our business partner, it was clear we needed to really look hard at replacing our existing z800 with a z9.
As we dug deeper into the costs of the potential upgrade, it became more apparent that this had a chance to be the most cost-saving processor upgrade we’d ever proposed. The numbers looked so good we were almost afraid to approach management with what we were seeing. It took another week of crunching the numbers before we made the pitch for the processor upgrade. We’ve emphasized the Intel server platform here and weren’t sure we could even get management to discuss a mainframe upgrade, but the bottom-line savings were hard to ignore:
• The potential for a $4,508 reduction in monthly IBM software costs
• A $120,000 reduction in maintenance the first year due to the warranty