7.6. Typical Metric Values
What, then, would be considered normal or typical values for these metrics? Of course, the best answer is given by Rule #1: “It depends.” Run-queue and load-average metrics have values that depend on the following, and it is not useful to provide a single rule of thumb for the values of these metrics.
Whether the application is interactive or batch
Whether the application is compute- or I/O-intensive
Whether the application is a server or client application
Speed of the CPU
Time in queue (the glance metric % Blocked on Pri is a measure of this)
One heuristic can be provided for the global system call rate. A rate of 3,000 syscalls per second per CPU may be indicative of a CPU bottleneck, although this heuristic will increase ...
Get HP-UX 11i Tuning and Performance now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
