March 2004
Intermediate to advanced
560 pages
14h 38m
English
Many tools provide useful data for dealing with CPU bottlenecks. Table 7-1 provides a summary.
| glance/gpm | MeasureWare | sar | top | vmstat | iostat | |
|---|---|---|---|---|---|---|
| user | X | X | X | X | X | X |
| nice | X | X | X | X | ||
| nasty | X | |||||
| real-time | X | X | ||||
| system | X | X | X | X | X | X |
| interrupt | X | X | X | |||
| context switch | X | X | SSYS | |||
| vfault | X | |||||
| trap | X | X | ||||
| wait I/O | X | |||||
| idle | ||||||
| run Queue | [*] | [*] | X | X | X | X |
| Load Average | X | X | X |
[*] Not shown by default; may be accessed through the Adviser.
Not all of these tools always provide accurate values for the various CPU data. In particular, sar, top, vmstat, and iostat don't always accurately account for interrupt time. In fact, they may count interrupt time as user time, idle time, or system time, depending on when the interrupt arrived. ...
Read now
Unlock full access