The article “The Performance Information Gap Paradox” (June/July Enterprise Tech Journal available at http://enterprisesystemsmedia.com/article/the-performance-information-gap-paradox) discussed the challenges of leveraging recent System z I/O technology enhancements without a solid grasp of how to optimize them and, more broadly, our diminished understanding of performance management in general. This article will continue that discussion and assume there was an application Service Level Agreement (SLA)/Service Level Objective (SLO) for transaction response time that wasn’t being met.
Was this caused by something in the I/O environment? This article will go through one of the key RMF reports used in mainframe I/O performance management/troubleshooting: SMF 74-1, the RMF Direct Access Device Activity report. This report contains response time information and information on the various components of response time. It can be used to further narrow down what may be the root cause of the problem and provide a good idea of what other RMF reports we should check. Figure 1 summarizes some of these reports and which components they’re used with.
Figure 2 shows an example of a Direct Access Device Activity report. This report shows little to no activity, but our purpose here is to discuss the various fields, their meaning, and what they measure.
The purpose of this report is to give information on I/O device use for all online devices you requested. This can be done by device number, device class, or volume serial number (VOLSER). The report helps with analysis of device performance, identification of bottlenecks caused by particular devices, and in overcoming inefficient resource use.
As CPU speeds increase, the I/O response time becomes the determining factor in the average transaction response time. I/O response time consists of two components: I/O service time and I/O wait time. We can use the following formula to illustrate this:
I/O Tr=I/O Ts + I/O Tw
The average I/O response time is one of the fields contained in Figure 2. We can get better response time by decreasing either I/O wait time (Tw) or I/O service time (Ts). Figure 3 further defines the components of I/O response time.