|
|
How is Response Indicator measured in VP for OpenVMS? |
|
|
A common way to measure system performance is to look at responsiveness. In ViewPoint for OpenVMS, this is known as the "Response Indicator." The Response Indicator has three basic parts, and its value is relative to the interval, the system-type, and the currently running workload on a the system of interest. (That is, a value of "8" on one system cannot be cross-referenced, extrapolated, and applied to another system.) Moreover, it is not an absolute measure like the commonly known measurement VUPS (Vax Unit of Processing), where one VUP is the baseline for CPU "horsepower."
The three basic parts that are referenced in the code to calculate the value are:
1. A compute-related factor
2. A disk-related factor
3. A pagefault factor
The first part is calculated by adding the # of computable processes plus the # of current processes and dividing by quantum:
(# of computable processes + # current processes)/quantum
The second part is calculated by getting the average response time for ALL disks by dividing the qlength per interval by the operation count per interval times the average operation count per disk:
(qlength/opcount per disk) * (average operation count/disk)
The third part adds the total hard page faults for reads and writes and divides this by the total faults system-wide:
(total # of hard page faults reads & writes)/(total pagefaults/system)
The Response Indicator is then calculated by plugging in the values from the formulas above and weighting them as follows:
Response Indicator = {((formula #1)*12) + ((formula #2)*8) + ((formula #3) * 3)}
Thus:
Response Indicator = {((formula #1)*12) + ((formula #2)*8) + ((formula #3) * 3)} |
|