Macro to Micro: Understanding z/OS Performance Moment by Moment
Project and Program:
Core Platform,
z/OS Systems Programming,
Performance and Capacity Management
Tags:
Proceedings ,
SHARE Kansas City 2024
Most customers report on most z/OS performance in 15-minute intervals, which is
fine for most use cases. But those 15-minute intervals gloss over the complexity
of what happens on time scales more relevant to the CPU: microseconds to
seconds. While performance analysts usually don’t need to worry about that
complexity, having a mental model of what’s going on at the micro scale can help
explain seemingly confusing results such as having TSO users complain about poor
response time when the reports indicate the machine is not very busy. Or why the
same work consumes different amounts of CPU at different times. Or why LPAR
configurations can impact performance. In this presentation Scott Chapman will
discuss performance from the 15-minute interval down to the microsecond level
and discuss how CPUs get virtualized and shared between LPARs and between
address spaces within LPARs. You’ll leave with a new appreciation for how the
micro level can impact the macro level.
Back to Proceedings File Library