A careful analysis performed for this Mexico-based banking and financialservices firm indicated that the routine batch processing of sort functions in the mainframe environment was costing about $1 million annually just in the consumption of CPU time.
Equally concerning was how the CPU time for sort jobs was encroaching on the 11 a.m. to 5 p.m. period when getting CPU time was critically important for vital business applications like transaction processing, CRM, and others. At stake were customer satisfaction, servicelevel agreements, and similar matters.
These two factors prompted a search for a DFSORT substitute, as DFSORT (IBM’s standard-issue mainframe sort utility) wasn’t providing what they needed. The bank’s mainframe IT group knew there were faster sort utilities out there, but which one was best? And how much difference would it make in terms of time saved and lower CPU usage charges?
Through its relationship with Devant Mexico, an IT and management consultancy in Mexico City, the bank became aware of Syncsort Inc. and of its flagship sort utility, Syncsort Elevate MFSort. The bank’s IT team was impressed enough with the joint Devant - Syncsort presentation of potential CPU-time savings that they asked Syncsort for a Proof of Concept (POC) exercise using portions of their actual banking workloads.
The POC confirmed expectations, so the customer licensed Syncsort Elevate MFSort and began testing it in their own mainframe environment. Two months later they began running Elevate MFSort in their production system.
Once Elevate MFSort was in the production environment:
Several months after Elevate MFSort implementation, Syncsort performed a workload optimization analysis for this customer over a 21-day period. Extrapolating results out to a full year from those 21 days, and using the industry-standard estimate of $0.12 per CPU second for mainframe sort processing, it was estimated that the customer was spending roughly $785,000 annually on its sort/copy/ merge tasks. This compares to the approximately $1 million before Elevate MFSort implementation.
While the comparison is rough — as it is based on extrapolating out to a year from a three-week time period — it suggests a better than 20% drop in CPU charges for sort processes following Elevate MFSort implementation.
Given this experience, the customer is considering adding Syncsort’s Elevate ZPSaver to the equation. Elevate ZPSaver extends Elevate MFSort by offloading a higher percentage of eligible workloads from the generalpurpose processor(s) to the zIIP processor(s) installed on IBM z Systems machines. Because the pricing on zIIP processors is not usage-based, as it is on the general-purpose processors, this can mean a further reduction in CPU usage costs.