• In addition to hardware costs to achieve specific capacities, most software licensing on z/OS systems is also capacity based. Due to these cost considerations, combined with the system’s inherent ability to utilize all CPU resources, z/OS systems tend to run at very high levels of system utilization. Reducing CPU requirements result in direct, near-term cost savings for customers.
  
  
• The z/OS environment has well established measurement metrics. Additionally, there are generally accepted performance ratings for zSeries systems. While there are several rating systems used by customers, most use MIPS (Millions of Instruction Per Second) as the unit of measure. These rating systems are not identical in their estimates of relative processing power among models, but the differences are typically minor. As a result, one can measure capacity requirements on one machine model and can make reasonable predictions of capacity requirements on other machine models.
  
  
• Depending on the methods used, the CPU time reported to a specific transfer task does not account for much of the total CPU time consumed by the file transfer operation. Uncovering and quantifying these hidden costs, as well as providing alternatives to these methods, can significantly improve mainframe critical areas of capacity planning and usage charge-back.

All benchmark runs used standalone processor environments with no other workloads present. To achieve the objective of measuring all CPU utilization associated with file transfer, both from the transfer task and any operating system overhead, we collected Resource Management Facility (RMF) data that recorded the total CPU seconds consumed by all work. The following formula calculates the CPU path length, expressed as millions of instructions per megabyte:

(( TCPU ) - ( ICPU )) X MIPS	=	Millions of Instructions
TRATE		per Megabyte

Where:

• TCPU = Total CPU seconds recorded during the period of file transfer
• ICPU = Measured CPU seconds when machine is idle for the equivalent period
• MIPS = Machine performance rating in Millions of Instructions Per Second
• TRATE = Transfer rate in megabytes per second

In addition to the CPU measurements, this report provides throughput measurements. Unlike the CPU measurements, which are applicable across environments, throughput data is specific to the measured environment. Variations in processor and disk speeds will influence results. This report presents this data only as an indicator of order of speeds, (fastest to slowest) among the benchmarked configurations.

Testing consisted of 14 runs for each configuration. Seven (7) runs pushed data from z/OS to a target system. Another seven runs pulled data from a source system to a z/OS target system. The seven runs varied logical record length (LRECL). The file block size (BLKSIZE) was set to half-track blocking of 3390 disk format, the most common blocking factor used in z/OS environments. The following table shows the combinations of LRECL and BLKSIZE used.

LRECL	BLKSIZE
80	27920
133	27930
2048	18432
4096	24576
8192	24576
16384	16384
27998	2799

Benchmark Environment

• IBM Model z9BC-R07, Capacity Level A02 (2-way, 50 MIPS)
• Eight (8) GB Main Storage
• FICON-II
• OSA Express 2 - GbE
• Storage –IBM DS 6800
• z/OS V1.8

AIX

• IBM 9131 Model 52A with 1.65 GHz Power 5+, 36MB L3 cache