2. PERFORMANCE REPORTING ANALYSIS › 2.7 Graphic Analysis › 2.7.5 Paging Graphs › 2.7.5.1 Response Time for Page Data Set Type
2.7.5.1 Response Time for Page Data Set Type
GRAPHIC RESULTS
This graphic analysis provides a multi-line plot of response
times for the three types of page data sets: Local, Common,
and PLPA (Pageable Link Pack Area).
ANALYSIS GUIDELINES
The response time is the average for all devices being used
for a page data set type.
At the DAYS level of summarization, you should look for
consistency. For any day that showed unusual degradation of
response time, you should determine whether this was due to a
periodic peak load that must be planned for in the future, or
whether it was due to some system problem or unusual load
that is unlikely to occur frequently.
There are essentially only four ways to respond to peak loads
and consequent increases in paging response times:
o Attempt to make a business case for shifting the
workload. Although a convincing business case cannot
often be successfully made, this can sometimes provide
temporary relief for acute paging response problems.
o Recommend the acquisition of additional memory. This
is the most effective and direct, but generally the
most costly solution.
o Add additional paging volumes or improve the
performance of the existing ones. This is generally
an attractive and not overly costly solution.
However, additional paging volumes cannot compensate
entirely for a gross overcommitment of processor
storage.
o Attempt to find an offending application and recommend
its redesign or other modification. This is often a
fruitless endeavor. However, you may occasionally
discover an application that has dramatically
overcommitted I/O buffers, is searching large arrays
in an inefficient manner, or is otherwise misusing
processor storage.
At higher summarization levels, especially when using the
MONTHS level time-span, you can look for trends. If paging
response time is gradually being degraded, this can give you
an early warning that you may need to take some corrective
action.
The detailed behavior of the paging subsystem can be further
analyzed using the Auxiliary Storage Management analysis
reports described in Section 2.2.