IBM OS/390 User Manual
⇒
Establishing performance specifications and policies.
•
Performance execution and measurement
⇒
Response time monitoring - what is the response time as seen by the
user?
user?
⇒
Availability monitoring - what
′
s broken?
⇒
Utilization monitoring - who are the biggest users? What is the service
times and queue lengths for key components and resources?
times and queue lengths for key components and resources?
⇒
Component delay monitoring - what are the bottlenecks?
⇒
Performance tuning - how can the resources be used effectively?
⇒
Tracking and control - what are the short term trends? alerts? long term
trends? how can I tailor the volumes and data flow to the needs and
resources of the system?
trends? how can I tailor the volumes and data flow to the needs and
resources of the system?
30.4.3 Methodology
Performance information must be available to view in real-time to get a snapshot
of system status, and to quickly address any problem or potential problem
situations. It must also be archived so that long term analysis can be carried out.
It is best to save this information to DASD or tape - printing and saving stacks of
performance information is a waste of time. It takes up too much space, you will
never find the information you need in a timely manner, and the information you
probably need at a given point of time will likely be in a report that has just been
discarded.
of system status, and to quickly address any problem or potential problem
situations. It must also be archived so that long term analysis can be carried out.
It is best to save this information to DASD or tape - printing and saving stacks of
performance information is a waste of time. It takes up too much space, you will
never find the information you need in a timely manner, and the information you
probably need at a given point of time will likely be in a report that has just been
discarded.
OS/390
′
s RMF subsystem will record system resource utilization information.
OS/390 SMF (Systems Management Facility) records contain RMF data and data
from other sources (JES, VTAM, NetView, NetView Performance Monitor) that
contains performance information. Many IBM and non-IBM performance products
that monitor specific resources can place their data into SMF records.
from other sources (JES, VTAM, NetView, NetView Performance Monitor) that
contains performance information. Many IBM and non-IBM performance products
that monitor specific resources can place their data into SMF records.
While users, I/O control clerks, operators or specialized jobs can capture the
business volume information, the use of performance and automation products
can help make this task less labor intensive. Understanding the relationship
between user transactions or batch jobs and business units of work will allow
business volumes to be directly derived from performance monitors, and
automation can be used to collect and format the data as needed.
business volume information, the use of performance and automation products
can help make this task less labor intensive. Understanding the relationship
between user transactions or batch jobs and business units of work will allow
business volumes to be directly derived from performance monitors, and
automation can be used to collect and format the data as needed.
Tracking performance data must also include online transaction response time,
DASD I/O service times, batch program elapsed times, and batch window start
and completion times. NetView Performance Monitor can measure and report
online transaction response time. RMF and SMF can provide data on DASD I/O
and batch performance. TME 10 Operations Planning and Control (TME 10 OPC)
schedules batch workloads and can provide data on batch program and batch
window elapsed times.
DASD I/O service times, batch program elapsed times, and batch window start
and completion times. NetView Performance Monitor can measure and report
online transaction response time. RMF and SMF can provide data on DASD I/O
and batch performance. TME 10 Operations Planning and Control (TME 10 OPC)
schedules batch workloads and can provide data on batch program and batch
window elapsed times.
Using a PC and a spreadsheet application to download and capture and report in
tables and graphs the critical performance variables over time is a very practical
way to start; junior systems programming personnel, senior operations
personnel, or help desk personnel can keep this information. A more robust
solution will be a performance reporting product such as IBM
tables and graphs the critical performance variables over time is a very practical
way to start; junior systems programming personnel, senior operations
personnel, or help desk personnel can keep this information. A more robust
solution will be a performance reporting product such as IBM
′
s TME 10
Performance Reporter; it can directly read many of the sources that contain
performance data, such as SMF, and produce short term and long term analysis
reports for use in capacity planning.
performance data, such as SMF, and produce short term and long term analysis
reports for use in capacity planning.
464
VSE to OS/390 Migration Workbook