Company: CybermationCustomer: JC PenneySubmitted by: MCC InternationalDate: March 2000Imagine running a computing operation today for less than your 1984 budget. Sounds impossible, but JCPenney is doing just that. With over 1,200 stores, the growing retail chain recently executed a well-led automation strategy. Not only are operating costs way down, but the company is handling an increasing workload while providing better computing services for its users — all of which puts JCPenney significantly ahead of its competition. “We have been right-sizing since 1970, restructuring every year to take advantage of technology,” says John Dratch, JCPenney’s director of data processing and technical support.
“By creating a control centre, as well as by right-sizing, we have now set the stage for the future.”
The key to JCPenney’s strategy was the restructuring of four large data centres responsible for running roughly 18 million jobs a year and staffed by 425 people. This distributed operating environment, with some 2,300 MIPS installed and six terabytes of storage, has now been centralised into a single control centre based in Plano, Texas, just north of Dallas.The idea behind the initiative was to separate the people from the hardware. At an estimated cost of $30-$35 million to rehouse the technology, moving the hardware didn’t make sense. Instead, it remained where it was: in four state-of-the-art, environmentally up-to-date and company-owned buildings in Reno, Lenexa, Columbus and Dallas.
By taking the people out of these environments and centralising them in a single Information Technology Control Center (ITCC) in Plano, JCPenney essentially created four unstaffed, lights-out hardware repositories.”We have a huge investment in legacy mainframe systems and there was no way we were going to throw away billions of lines of code and recreate applications,” says Dratch, adding that he believes the company’s long term commitment to its mainframe environment has provided the kind of disciplines which should ease the transition to client/server distributed technology.For example, aspects such as the help desk, network management, performance monitoring, capacity planning, problem solving and change management pertain to both mainframe and client/server environments. Reorganisation of files, disaster planning, backup and recovery all need to take place as well. As long as JCPenney has network connectivity to devices wherever they are located, it can apply the same system management disciplines that worked for its mainframe environment to any future client/server implementations.
The thrust of JCPenney’s automation strategy was to let the computers manage themselves. As it moved ahead, the retail organisation realised this would entail replacing some of its existing software with new products and tools in order to consolidate and reduce work effort.
RIGHT TOOLS FOR THE JOB
To do so, it identified two strategic issues: scheduling and automation. The scheduling process at JCPenney is not only complex, but handles large enough volumes to demand a production management system, not just a scheduler. The system needs to know when jobs are supposed to start and end, how long they should take, and what dependencies exist. Then, if workload doesn’t go exactly according to schedule, the system needs to take appropriate actions automatically, or it might involve opening an incident report to flag a problem that requires human attention.
In short, scheduling and automation functions need to work together as a cohesive pair.To meet these objectives, JCPenney replaced its old scheduling system with ESP Workload Manager, a job scheduling and workload management system from Toronto-based Cybermation, and brought in Automate, an automation product from Legent Corp. in Herndon, Virginia.”These two products work hand-in-hand with each other and feed off of each other,” points out Dratch. He adds that a third product, IBM Corp’s Info/Sys, also interfaces with ESP and Automate to provide the incident report (IR) problem management process.
“We look at these software products as one entity,” he says. “Each one by itself is good, but the sum of the three together is very valuable and strategic for us.”In the new control centre, one manager and four people are responsible for writing rules for automation; another manager with four people manage the database with the schedules in it; and, another team of five are involved with the change control process. In total, staff has been reduced by 75 per cent to 135 people including administrative staff; Help Desk, which handles 35,000 calls a month; console operations, which handles exceptions not covered by automation; problem solvers; and performance management staff.Together with these reductions, the entire system has been streamlined and is now highly effective. For example, with the old scheduling system, costly production delays of two to three hours occurred whenever schedules were being built.
As the company migrated to ESP Workload Manager, these delays disappeared.Because the philosophy behind ESP Workload Manager is to eliminate all redundant work, in most situations it was possible to come up with a single coherent method to do the same work in multiple places.”Literally thousands of lines of code which we had in the database to describe schedules under the old system became only 10 lines in the new one,” remembers Julie Norris, who was then ITCC manager of scheduling. “There was effort in converting the old to the new, but in going forward, the management of schedules has been improved a hundred-fold.”Another benefit to centralising control of the four data centres is standardisation of business practices. “We had four data centres and we swore we all did things the same way,” recalls Barry Taylor who manages the ITCC.
“This migration forced us to take another look and our eyes were really opened — we weren’t doing things the same way at all.”In addition to helping create standards, the ESP software also helped consolidate the number of different job definitions. For example, over 60,000 job definitions have now been reduced to 15,000. “Where we used to have 626 jobs in one application due to repetitions, we now have only 10,” points out Norris, “ESP just runs them over and over again.”ESP also enables a more efficient operation of JCPenney’s multiple credit service centres, all running the same schedule.
Using the symbolic processing available in ESP, they could code just one schedule for all the centres, reducing 832 jobs to 94.
MIGRATION NOT CONVERSION
Jim Blanton, JCPenney manager of strategic planning for operations, points out: “The one thing we did right was doing a migration rather than a conversion.”According to Blanton, the team felt using a conversion tool would have brought all of their old “sins” and inefficiencies into the new workload management system. “One of the primary rules of automation is don’t automate the old, erroneous ways of doing things,” he says. “The trouble with doing a quick and dirty conversion is that although it’s quick, you’ve got a dirty end product.
“Migrating to a new approach and new software, on the other hand, has resulted in a slick operation that eliminates those inefficiencies. One area where this is most noticeable is in problem detection.
In its new system, JCPenney uses what it calls “Probes” which continuously go into the system to provide updates every 15 seconds. Since ESP and Automate “talk” to each other, it is possible to display on huge screens whether or not things are going the way they should for all system components. If they aren’t, a red light indicates the problem area.With everyone viewing a problem simultaneously and reacting accordingly, a lot of unnecessary communication has been eliminated.
Before, such communication had to take place by word of mouth, by telephone or person-to-person. Now, the probes flag problems as soon as they occur.For JCPenney, automating has meant eliminating the time-consuming, manual tasks that people do. If there is a failure, the system makes a phone call to the appropriate systems analyst, and manages the escalation process automatically.”Our job is to provide services to the company at the most effective and economical cost possible,” says Dratch. “This means getting as much of this activity done seamlessly by the computer as we can.
“One area where the most significant impact has been made is the automatic opening of incident reports. Initially, the plan was to have five people monitoring problems per day, per shift. When the centre opened, Dratch says he found he needed at least three more people and even then they were missing an occasional IR.”Whichever one they missed happened to be the most important one of the day,” he adds. “When we got our automated IR process going, we saw a terrific reduction in people needed and a big improvement in accuracy.” With ESP Workload Manager, when a batch job abends, an IR is automatically opened.
ESP Workload Manager collects all of the data needed and passes it to the person doing the problem-solving. For example, it identifies the data set name where the job is located.Here’s where the integration with Automate comes into play. Once the IR is written, ESP Workload Manager tells Automate the JES number, the job name and IR number. Automate can then pull the sysout automatically, updating the Info/Sys IR so the problem solver knows where the sysout is.Analysts or problem solvers can be notified of a failure at any time, no matter where they are.
The system places the IR in an analyst work pool and then automatically pages the analyst, providing text to describe the nature of the problem. If there is no response within 15 minutes, the system pages a second time. Then, if there’s still no response, the IR automatically escalates to the Analyst Manager so nothing is lost.
AUTOMATION BRINGS FINANCIAL SAVINGS
Blanton points out that the opening time for IRs is now less than one-tenth of a second, resulting in a tremendous productivity increase. In the old manual system, operations had to look at hundreds of messages flowing by and hopefully notice when something was wrong.
They would then open an IR by calling someone, posting an IR in the system and getting the documentation together — all of which could have taken up to half-an-hour.Fixing problems may take as much time as it did before, but the fact that we now have the information, know about it right away and get the resolution out there quickly, really reduces the outage,” says Taylor.Now, ESP Workload Manager traps the list job output and shows which step it went down in. Everything is captured automatically, as opposed to having someone search the logs looking for some kind of error message.Reducing the 20- to 30-minute information gathering time down to a matter of seconds using ESP Workload Manager and then Automate to get the sysout has not only created huge time savings, but has had a significant financial impact as well.
Dratch and his key managers aim to provide a competitive advantage to the company through low-cost, effective computer services. The automation strategy has led to a net result of $11.5 million (U.S.) in payroll savings alone for 1994.
Dratch and his team have a strong appreciation for JCPenney’s business requirements. They see themselves as a support organisation and if they do a poor job, they know all of the company’s businesses — whether stores, catalogues, credit or insurance — will suffer.So while they forge ahead, continuing to seek ways to maintain JCPenney’s competitive advantage, internally they continue to refine and advance their automation processes. In all of this, one thing is clear to Dratch: “We wouldn’t be as far along in our automation strategy if it hadn’t been for the success of ESP Workload Manager.”JCPenney’s external auditors told them they had a two-year advantage over their competitors because of what they have done with their automation strategies, and the corresponding reduction in cost of operations. ESP Workload Manager is providing high value to JCPenney and forms a highly strategic element of their automation strategy.
Their forward-looking approach will help to ensure that JCPenney maintains a competitive advantage in the years to come.