The Projects

62% SAVINGS IN COMPUTER RUN-TIME.

A Fortune 25 West Coast Bank had various performance problems one of which was erroneous CIsize errors. Upon an extensive analysis, TURBO™, my analysis tool, found the problem that when implemented; the senior tech on the job wrote us the following email:

“Yes we installed and we did see great results the fiche processing that normally starts the 3rd business day of the month and has over 45,000 jobs ran in less than 9 hours – normal time is 24 hours. I see no need to take the time to send new files. We have tracked the improvements. Thanks for the recommendation this one worked out nicely.”

This saved 62% of the run time allowing the remaining hours (16) not processing this application to process many more jobs, which usually had to wait 24 hours for the Fiche jobs to finish.  What is that worth to a 24/7 data center??

MAINFRAME PERFORMANCE TUNING. MAINFRAME COST REDUCTION.

A client ran a very large Airline Reservation system and wanted to reduce the CPU charges they paid an Outsourcer and reduce the very long run time of a monthly job.

The monthly job ran for 14 days each month. It was a CICS IMS Data Base containing 200,000,000 records running a Month End Edit, Update and Reporting process.

Obviously, a 14-day run time was unacceptable but the outsourcer could find nothing wrong. A consultant who worked there knew Ralph and recommended they contact him.

Ralph did an in-depth analysis using his TurboTune tools and found many run-time improvements.  Every month the Job ran in an 8 Meg Region in low priority.

His findings were tested and implemented into production. The next month the job ran in 18 hours instead of 14 days, and the CPU was the same. It was in the same low priority and 8 Meg Region.

The client was delighted about the run time. The outsourcer was delighted and got the use of an 8 Meg Region for an extra 13 days and 6 hours to use for other clients.

NOTE: Run Time must be taken into consideration when calculating savings.

39% SAVINGS IN COMPUTER RUN-TIME.

A large New York based Financial Services firm was in the process of extensive reprogramming of their financial applications (Y2K). A major Accounting/IT Consulting firm was doing the programming work and the data processing was outsourced to the computer manufacturer.

As the project progressed, it was becoming apparent that the programming changes were so extensive that the normal workday was inadequate to do all the necessary testing. Testing was overflowing to weekends at a time-and-a-half/double time billing rate.

This not only extended the programmers time but also the computer operators had to be included in the extended time frame. The costs were becoming unmanageable! Neither the outsourcer nor the programming contractor could solve this problem.

The CIO of the firm had heard of Turbo™, my analysis software and called me. I was asked to improve mainframe efficiency by shortening on-line transaction response time and reducing batch consumption and run time especially in the batch production cycle.

I ran my Turbo Analysis and reduced consumption by an average of 39%, this allowed testing to resume during normal work hours and delayed a hardware upgrade for 18 months.

40% REDUCTION of EXCESSIVE STORAGE.

This New York based bank wished to consolidate 17 Country’s data centers into one in Singapore. The problem was there was no extra room to hold all of this data.

And the Sydney data center was experiencing severe online and batch window problems.  The CTO called up their New Jersey data center to ask for help and they recommended Ralph and his TURBOTUNE™ suite of performance products.

Ralph arrived in Sydney and the vendor of the Systematics package was also there but they could not solve the problem. After an overnight analysis, Ralph discovered 800 file improvements. The fixes were implemented and solved all performance problems.

The DASD footprint was reduced by 40% allowing the bank to migrate all data centers to Singapore saving millions of dollars. The service was what New Jersey said it would be—exemplary.

CONVERTING DATABASES AND SAVING RESOURCES.

After a very large New York based bank purchased the largest Student Loan Company in the United States, all files had to be converted into the banks database going from an IDMS ADSO database into an AMS vendor package. The Student Loan Company was processing 10,000,000 student loans per year.

The problem was an exceptionally long run time in the conversion to the new system preventing the system conversion to take place. The test runs were taking over a week on the largest mainframe that IBM made at the time. This was not acceptable. I ran my TurboTune analysis and recommended performance improvements. After the identified improvements were implemented the conversion ran successfully over the Fourth of July weekend in less than 24 hours.

IMPROVING PERFORMANCE AND LOWERING COSTS.

This very large Insurance Company had performance issues and after asking several of their vendors to study the problem, the consensus was that there were still lots of low hanging fruit available but no one seemed to offer any solutions.

The environment consisted of CICS, IMS, and Vsam vendor products.  Ralph was called in after the client saw the TurboTune website and Ralph did an analysis using his TurboTune tools. The analysis indicated that there was another 15% improvement available to save MIPS and DASD. The improvements were incorporated into the data center with great success. Ralph is still an active vendor doing various tuning projects.

MERGING TWO VERY LARGE BANKS. $20,000,000 Savings in MIPS

When these two banks merged, they needed to improve performance and save DASD but there was no internal group that was capable of doing this. Ralph was retained and after his extensive analysis, discovered that there were over 1,500,000 files and supplied improvements for 125,000.

Ralph met with application managers to assist in implementation and to help systems technicians create internal tools to identify problems on an on-going basis. To date, the combined bank has saved over $20,000,000 in MIPS, 10,000 hours in Batch processing time and 3,000,000,000,000 bytes of DASD plus many other performance options and parameters were identified.

BATCH RUN TIME GREATLY REDUCED.

The main problem with this Health Insurance Company was that they could not “consistently” complete their nightly batch work before it was time to bring up the CICS regions in the morning. To quote the manager who ran this application:

“The SMF analysis, however, allowed me to report the improvements more scientifically.  What I was most pleased about was that the amount of “cache hits” we saw when accessing our data increased substantially and our CICS response times improved. 

The higher your I/O content, the more substantial your improvement should be.  I didn’t have to show an ROI as it relates to $$$, as this was a performance-based initiative.  The SMF type42 records were the data source for my reporting.

We saw improvement here that exceeded what Ralph had guaranteed.  At the conclusion of the effort, I wrote a process that analyzed our SMF data to report on performance-related improvements that were realized. 

I put this into a spreadsheet and graphed it for a presentation at our (at that time) Technical Architecture meeting.  My boss was very satisfied with the improvements we achieved (as was I).”

REMOVING EXTRA PRODUCTION PERSONNEL.

THE MORTGAGE SYSTEM of this New York Bank was experiencing production slowdowns at a rate of 540 per month causing 1,500 employees and temps to stop work for 7 to 12 minutes several times a day and the nightly batch cycle ran into the daytime hours so the CICS on-line system could not come up on time in the morning. Ralph was brought in and analyzed the system using his TurboTune tools.

An analysis was performed using TurboTune tools, 20 year old parameters and settings were discovered.  Performance improvements were identified and after implementation, there were no more production slowdowns. The very expensive Saturday work force was eliminated as was the rent, extra computer resources, overtime, and all extra data center help.

IMPROVING A MAILING LABEL SYSTEM COST REDUCTION. Run Time Reduced from 18 Hours to 1 Hour.

A professional society ran a mailing label job each month for 300,000 members to mail a professional journal. The programs were written in COBOL and the monthly run took 18 hours to run at a service bureau cost a lot of fees. Ralph ran his Turbo COBOL Analysis tool and discovered an improvement to the programs. The programs were changed and recompiled. From then on the monthly processing took only one hour.

CICS ONLINE PRODUCTION SLOWDOWN.

One of the largest banks in the United States had a very serious production slowdown problem. At many random times a day an entire LPAR would go into slow motion and no one could find the cause. Ralph was called and did an analysis using his Turbo CICS Analyzer. This identified a vendor file and a vendor program that was causing a 60% wait on catalog condition every time a query was submitted. The vendor was contacted, the program was corrected and the slowdowns were gone.

MAKES A LONG RUNNING PAYROLL MORE TIMELY. From 2 Days to 20 Minutes.

A PRODUCTION Payroll system program was running for 2 days and Payroll was not being met on time. Ralph was asked to see if anything could be done to speed up the process. Ralph ran his TURBO COBOL ANALYZER and found a segmented file was being created in the incorrect sequence. The program was corrected and recompiled. No program logic was changed. The program ran in 20 minutes thereafter.

SAVING MILLIONS IN PRE OUTSOURCING DOLLARS. $540,000.00 Savings.

One of the largest banks in the United States wanted to outsource their VISA CARD processing. They wanted to reduce their costs prior to outsourcing. The annual cost to run the in-house processing was $ 66,000,000.00 a year.

Ralph ran a TURBOTUNE assessment and discovered 2,000 files where CPU could be reduced and speed of processing could be vastly improved. A project was begun to implement improvements.

The very first month the bank reduced CPU costs by $ 540,000.00 and only had completed 20% of the improvements. A very happy client when they outsourced their system.

MAJOR BANK SPEEDS POSTING – IMPROVES REPORTING.

One of the largest banks in the United States had a CICS online General Ledger System that would collect 2,500,000 detail transactions on line each day and edit and post them over night. Each morning they would get reports showing the invalid transactions and fix them for the following nights processing.

Millions of dollars of transactions were being delayed a day costing the bank a great deal of interest and under reporting the banks condition to The Federal Reserve every Tuesday.  Ralph was called and an analysis was run using TurboTune and all of the files were identified. Ralph built a shadow system that was implemented that allowed the bank to capture all of the errors and correct them the same day. Problem solved.

SECOND LARGEST BANK MERGER IN U.S. HISTORY. Batch Run Time Reduced from 24 Hours to 5 Hours.

Two very large banks were merging. They needed to run a combined General Ledger using a new combined Chart of Accounts. The new bank would use one of the existing bank’s Vendor General Ledger Application. The new General Ledger would accept duel transactions from both old systems and produce both old and new daily reports.

During testing it was found that the overnight batch process was taking 24 hours to run and that meant that online could only be used every other day. This was not at all acceptable. The vendor was called and studied the problem.

The vendor sent in their very best senior systems people and they could find nothing that would improve the speed of processing. Ralph was in a nearby cubicle listening to the discussions and begged the General Ledger Manager to take a look at the problem.

Ralph ran his TurboTune process and discovered hundreds of performance improvements. Ralph’s recommendations were tested and moved to the new system. With Ralph’s improvements the batch process now ran in 5 hours each night and CICS could come up on time every morning with time to spare.

After the merger was complete and the new system was in place, the cost accounting system showed that the new system was costing the same as the old one bank system even though the number of records being processed had quadrupled.