Writer instructions | |
---|---|
Page title | For most spaces, this page must be titled Space announcements. For spaces with localized content, this page must be titled Space announcements l10n. |
Purpose | Provide an announcement banner on every page of your space. |
Location | Move this page outside of your home branch. |
Guidelines |
CPU Load (CPULoad )
Description
This parameter displays the 1-minute load average from the uptime command. Load average is the average number of processes in the kernel's run queue during an interval (1 minute in this case).
The data from this parameter is often more useful when looked at in combination with other parameters (such as CPUCpuUtil ). A high load average along with low CPU utilization can indicate problems (most likely in the I/O subsystem).
For most UNIX systems, the load average from the uptime command is not a pure measure of CPU use because it includes processes waiting for disk and network I/O. However, used with other parameters, CPULoad can help to identify several problems as shown in the Recommendations section below.
Recommendations
If your system appears to be slow, run the uptime command to check the 5-minute and 15-minute load average. (Consult your system documentation for information on the uptime command. Some systems have a different implementation of this command.) If the 15-minute load average is higher than the 1- or 5-minute load averages, then the average is falling, and the problem may resolve itself shortly.
A consistently high load average can indicate performance bottlenecks in one or more of the following areas:
- Memory: If the number of page faults (MEMPFault) and/or page-outs (MEMPageOut) is high and your load average is high, your system probably needs additional memory to improve performance.
- Disk/controllers: If your load average is high and the CPU has excessive waits for I/O operations (CPUWio), then your system can probably benefit by the addition of a controller and/or disk drive.
- CPU: If the first two areas are normal, then you may have reached the limit of your system's resources. The following suggestions may help:
- If your CPU load is high and a particular process that is important to you is not executing quickly enough, you can use the renice command to increase the process' priority. Refer to your system's documentation for information on the renice command.
- If you have batch processes running, you can schedule some jobs to run during non-peak times.
- If the previous bulleted items do not help, you may need to increase your system's CPU capacity.
Because each UNIX system may have vastly different characteristics, no range thresholds have been set by default. With the combination of your experience with your system and your system documentation, you can set ranges that are useful.
Default parameter attributes
Attribute | Default value |
---|---|
Application class | CPU |
Command type | not applicable |
Platform | all |
Icon style | graph |
Unit | load factor, which is the average number of processes in the kernel's run queue during a 1-minute interval |
Border range | undefined |
Alarm1 range | undefined |
Alarm2 range | undefined |
Scheduling (poll time) | inherits poll time of collector |
Active at installation | yes |
Parameter type | consumer |
Value set by |
BMC ProactiveNet Performance Management properties
Property | Default value |
---|---|
Monitor type | CPU |
Key Performance Indicator | No |
Monitor for abnormalities | No |
Graph by default | No |
Availability | No |
Response time | No |
Normal distribution | Yes |
Statistical | Yes |