HP 3000 Manuals

CPU Detail Screen [ HP GlancePlus User's Manual (for MPE/iX Systems) ] MPE/iX 5.0 Documentation


HP GlancePlus User's Manual (for MPE/iX Systems)

CPU Detail Screen 

If you choose the C command for the CPU Detail screen, you can view
metrics that are specific to CPU usage.  In addition to the Program
Banner Line and Global Bars, this screen will show CPU usage by the
dispatcher, memory manager, system, and batch processes, as well as the
time spent in various states such as paused, idle, or in a queue.  The
detailed data section of the display has four components:

   *   Type of Utilization.

   *   Queue Distribution.

   *   Miscellaneous Performance Metrics.

   *   Top CPU Consumer.

	       Click here to view figure.
          Figure 5-15.  CPU Detail Screen 

Type of Utilization 

This uppermost section of the detail area lists the percentage of total
time used for different system resources and system purposes.

	       Click here to view figure.
          Figure 5-16.  CPU Detail Screen--Type of Utilization 

These system resources and system purposes include the following:

MemMgr       This line shows the time during the current
             interval that the CPU was processing certain
             Memory Manager activities, such as fetches and
             swap ins.

             If memory manager CPU utilization is
             consistently higher than 8%, the system might be
             experiencing a main memory bottleneck.

System       This line shows the time during the current
             interval that the CPU spent on system processes
             such as NS services, spoolers, and the Loader.

Session      The Session line shows the time the CPU spent on
             interactive processes during the current
             interval.

             If more than 70% of the CPU time was spent,
             consistently, on interactive processes, the
             system might be experiencing a CPU bottleneck
             condition.

ICS          The Interrupt Control Stack (ICS) line shows the
             time spent on interrupt activity during the
             current interval.  The ICS value is derived by
             subtracting the sum of all other CPU states
             during the interval from 100%.

Dispatch     This line shows the time the CPU spent on MPE/iX
             Dispatcher activity.

Batch        This line shows the time the CPU spent
             processing batch jobs during the current
             interval.

          Table 5-1.  (cont.) 

Pause        This line shows the time the CPU was paused for
             the disc during the current interval.

             If more than 25% of the time is spent,
             consistently, in a paused condition, the system
             might be experiencing a disc bottleneck
             condition.

Idle         This line shows the time the CPU was idle during
             the current interval.

Queue Distribution 

The Queue % line shows the active CPU cycles consumed during the current
interval distributed by queue.  The percentage for each queue is shown at
the right.

	       Click here to view figure.
          Figure 5-17.  CPU Detail Screen--Queue Distribution 

Miscellaneous Performance Metrics 

The bottom section of the CPU Detail screen shows various Global CPU
performance metrics measured during the current interval.

	       Click here to view figure.
          Figure 5-18.  CPU Detail Screen--Miscellaneous Metrics 

Switches to CMhis shows the rate of switching from Native
             Mode to Compatibility Mode.  [Bar Scale:  0 to
             500 switches per second.  The number to the
             right of the bar shows the actual switch rate
             per second up to a maximum of 9,999.]

Switches to NMhis shows the rate of switching from
             Compatibility Mode to Native Mode.  [Bar Scale:
             0 to 500 switches per second.  The number to the
             right of the bar shows the actual switch rate
             per second up to a maximum of 9,999.]


NOTE Excessive switching may indicate CPU saturation. While switching rates depend on machine size, you combine the rates for the two modes and use table 5-2 as a guideline for determining whether excessive switching is occurring. If it is, you should consider a Native Mode migration.
Table 5-2. Guidelines for Switches to CM or NM ------------------------------------------------------------------------------------------- | | | Machine Threshold Rate | | Model (switches/second) | | | ------------------------------------------------------------------------------------------- | | | S/920 approx 75 | | | | S/925 approx 120 | | | | S/922 approx 150 | | | | S/932 approx 225 | | | | S/935 approx 270 | | | | S/950 approx 300 | | | | S/955,917,927,937,947 approx 440 | | | | S/948,949 approx 500 | | | | S/960 approx 630 | | | | S/958 approx 650 | | | | S/957 approx 720 | | | | S/980-100,967 approx 1000 | | | | S/977 approx 1170 | | | | S/980-200 approx 1500 | | | | S/980-300 approx 1830 | | | ------------------------------------------------------------------------------------------- CSI243sIX560CSI32s Interval CM % This shows the percentage of time the CPU was in Compatibility Mode between switches. Interval Processhis shows the rate of process completions for Completions the current interval (number of processes completed per second). This provides information about the workload characteristics of the machine. A high rate of process completions, for instance, may indicate a CPU bottleneck due to excessive or improper process handling. A sustained rate greater than 0.50 may be considered excessive. Current This shows the number of processes waiting for Ready Queue the CPU. [Scale: 0 to 25 processes.] Maximum This shows the maximum number of processes Ready Queue waiting for the CPU since GlancePlus was started or since the MPE/iX Measurement Interface was enabled. [Scale: 0 to 25 processes.] Launch Rate This shows the rate of launches during the current time interval (number of launches per second). A launch occurs every time the MPE/iX Dispatcher gives a process control of the CPU. CQ SAQ This is the System Average Quantum (SAQ) for the C queue (CQ). This value represents the current time limit (in milliseconds) during which a process in C queue can retain use of the CPU. The SAQ value, which is recalculated continuously by the dispatcher, must be a value between the minimum (MIN) and maximum (MAX) parameters of the current CQ setting. Interval This is the rate that files are being opened File Open Rate during the current interval (number of files opened per second). The File Open Rate can be used to judge the workload characteristics of the system. A sustained rate greater than 35 files opened per second might indicate a CPU bottleneck condition or, possibly, poor application design. Top CPU Consumer In a single line near the bottom of the CPU Detail screen, GlancePlus names the process that is consuming the most CPU resources. Click here to view figure. Figure 5-19. CPU Detail Screen--Top CPU Consumer The percentage represents the amount of CPU resource that the process used during the last interval. The screen lists either the program name or the MPE/iX Command Interpreter (CI) command last issued. If an MPE/iX system process is the top consumer, it is identified by the program name or by the text "running program on behalf of MPE/iX." If you wish to analyze that process in more detail, you can use the > command to take you to the Process screen for that process identification number (PIN).
NOTE If you do not have System Manager (SM) capability, you will see the letters "CI" instead of the command string text.


MPE/iX 5.0 Documentation