Performance Intelligence > Clients > Success Stories > LeumiCard

LeumiCard improves computing infrastructure efficiency using Performance Intelligence

Leumi Card is a leading credit card company in Israel, that issues cards, offers card processing services and payment solutions. Leumi Card, a fully owned subsidiary of Bank Leumi, issues MasterCard and Visa cards and processes payments for businesses. The company issued more than a million cards and serves about 34,000 businesses in Israel.

Leumi Card is a leader in the development of new services and advanced payment services – adding value to its customers. Among the services and products that were developed are the “personal credit” Visa card, direct deposit (to transfer funds from Leumi Card accounts to any bank or credit card in Israel), Visa Y card – a pre-paid credit card targeted at teenagers, “Verified by Visa” payment service (secure Internet payment using Visa International standards) and more.

The company invests in building the marketing and operation infrastructure in order to provide the best possible service to it’s customers. The computing infrastructure includes IBM mainframes (zOS), Unix and Windows computers. The company invested millions of dollars in advanced information processing systems.

The business challenge

For Leumi Card, processing the end of the business day is done on the mainframe in batch. If the batch execution does not end on time -  some of the data might not be available for the next business day and the company's operations can suffer. Batch jobs must end before the start of the new business day.

Mr. Rafi Cohen, CTO:

"We have about a 1,000 jobs executing every night. The nights are not identical - some jobs execute only on specific days while others execute differently and consume different resources on different nights. Having done our implementation planning we know what the critical paths are. There are many modules and each one of them has a critical path. Until all the tasks of a critical path are finished - the specific module can not be started for service. Among the modules: Businesses, card holders and data capture. Every module uses hundreds of jobs - some of them belong to the critical path and others work in parallel to the online system executing in full service mode. Our business challenge is to shorten as much as possible the critical paths. Our technical challenge is to get the management information and analyze it to truly understand our performance situation. Given the amount of data and variables we can not use a pre-defined set of reports - we need to apply business intelligence techniques to analyze, understand and manage our performance data".

The business Advantage

Shortening the Batch window will increase the service level. It will also improve the computing infrastructure efficiency and will allow hardware upgrades to be postponed - or even scaled down.

The Solution

Leumi Card implemented the batch component of PIReporter for zOS with help from Performance Intelligence, Inc. The product allows the user to define information requests quickly and easily by defining parameter values. The data is provided (based on need) from a data mart hosted on a Windows server or from the original performance data on the Mainframe. Using a Windows based data mart enables the analysis of data without adding to the mainframe's workload. In addition - the original performance data can be used on the mainframe if the need arises.

Mr. Rafi Cohen describes what is done:

"We define priorities for the critical paths using WLM. It is important for us to understand what jobs execute in parallel to the critical processes. We use PIReporter to understand what jobs execute during the critical paths execution. We know the application and can give priorities to optimize the critical paths of the different modules. To shorten the critical path we need to know where to concentrate our efforts and identify the problematic jobs. We use different performance indicators to identify problem areas - an example can be CPU to I/O ratio. We can not, manually, inspect 1,000 jobs daily. We needed a tool that can point us to the critical jobs. Once we identify these critical jobs - we use PIReporter to inspect and analyze the different performance indicators to find problem spots - Elapsed time, CPU Time, I/O and the ratios between them.

We looked for a tool that can provide access to important performance measures: date and time of execution start, Elapsed, CPU and I/O information with detailed information for every step and summarization by job. We needed a tool that allowed us to define and execute ad-hoc reports quickly and efficiently so we could optimize and shorten critical paths and identify jobs that we could improve to remove in-efficiencies.

For example: We found a Cobol job that consumed a lot of CPU but needed very little I/O. Once we identified this job we used our performance Monitor to pinpoint where CPU resources were wasted. Apparently - we had a piece of code that used INITIALIZE to a very large memory block inefficiently. We fixed it and CPU consumption was halved. In another example we noticed a job with very large I/O (EXCP). If we see that I/O consumption is very high we need to check the buffers. In VSAM this is done external to the program. We optimized our VSAM and noticed astonishing performance improvement: 85% for I/O and 45% for Elapsed (An hour taken out of our critical path). We found out that once we were able to shorten our critical paths we were able to identify other jobs that can be optimized and improve our performance further".

Conclusion

Mr. Rafi Cohen concludes:

"Optimal management of Batch in a production environment is not a one-time ordeal. The ability to continuously obtain and analyze operational information that is not  directly connected to the daily operations brought substantial, permanent improvement to the efficiency and utilization of computing infrastructure at Leumi Card. PIReporter is now a standard part of the management infrastructure of the organization. Every month we use it to create comparison reports that can be shown to upper management and used to explain the changes in resources utilization and their causes".