|
|
|
The following performance metrics are monitored by the SystemEDGE agent.
The total active real memory (kilobytes) sampled over a fixed period. The sampling interval is system-dependent. In SunOS, the kernel updates this parameter once every seconds. Memory is active when a process running in memory actually uses it. This value should be less than or equal to memInUse. A high value can indicate a system with insufficient memory.
The total number of 'ticks' spent by all CPUs in Idle mode. This variable is not supported by NT. This value may wrap 32-bit counter.
The percentage of time (over the sample period) for which the CPUs of the system were idle.
The total number of 'ticks' spent by all CPUs in Kernel or system mode. This value may wrap a 32-bit counter.
The percentage of time (over the sample period) for which the CPUs of the system were executing the kernel or operating system.
The total number of 'ticks' spent by all CPUs in User mode. This value may wrap 32-bit counter.
The percentage of time (over the sample period) for which the CPUs of the system were executing in user mode.
The load average over the last 15 minutes. The load average represents the average number of jobs in the run queue over the specified time range. The value reported is the load average multiplied by 100. For example, if the value 50 represents a load average of 0.50.
The load average over the last 1 minute. The load average represents the average number of jobs in the run queue over the specified time range. The value reported is the load average multiplied by 100. For example, if the value 50 represents a load average of 0.50.
The load average over the last 5 minutes. The load average represents the average number of jobs in the run queue over the specified time range. The value reported is the load average multiplied by 100. For example, if the value 50 represents a load average of 0.50.
The percentage of the system's active memory in use. The value of this object can be approximated by dividing activeMem(10) by memory(system.3) and converting to a percentage. Because virtual memory allows systems to operate with a higher memCapacity value, this value alone should not be used to judge how loaded a system is. Managers should also monitor swapCapacity.
The amount of system's memory in use.
The percentage of the system's memory in use. The value of this object can be approximated by dividing memInUse(9) by memory(system.3) and converting to a percentage. Because virtual memory allows systems to operate with a higher memCapacity value, this value alone should not be used to judge how loaded a system is. Managers should also monitor swapCapacity.
The systemwide total number of open files.
The total number of page faults that have occurred since the kernel was last initialized. A page fault occurs when a running process attempts to access a virtual memory page that is not currently in physical memory. This attempt results in a hardware page fault. A high rage of page faults can indicate an overloaded system or one with insufficient memory.
The total number of pages that have been paged in since the kernel was last initialized. Page-in is an operation performed by the virtual memory system in which the contents of a page are read from secondary storage. A high rate of change can indicate an overloaded system.
The total number of pages that have been paged out since the last scan. Page-out is an operation performed by the virtual memory system in which the contents of a page are written to secondary storage. A high rate of change can indicate an overloaded system.
The number of processes table slots currently allocated and in use. On older BSD-based systems, this metric is important because it places an upper bound on the total number of processes that may run on the system at any one time.
The total number of context switches that have occurred since the kernel was last initialized. A context switch occurs each time a process gives up the CPU and another takes its place. This counter reflects the level of system activity; a high rate of context switching is indicative of the system load.
The total number of system calls that have occurred since the kernel was last initialized. System calls occur when a process calls a function which must execute in kernel or privileged mode. It is an indicator of the system load and can be higher on servers or computers shared by many users.
Current number of zombie processes found on the system. This MIB object is found by adding up all processes with a processStateStr of zombie. The lower the number the better. Supported in SystemEDGE 4.1 and higher.
The length of the scheduler's run queue sampled over a fixed period. The sampling interval is system-dependent. In SunOS, the kernel updates this parameter once every 5 seconds. The load average of the system is computed from the value of runQLen; the load average is defined as the average runQLen value over some interval.
The percentage of the system's total swap in use. The value of this object can be computed by dividing swapInUse(13) by totalSwapSpace(29) and multiplying by 100. It is included here for convenient polling, monitoring, and history sampling. This object is an aggregate value over all the swap partitions and areas.
The amount of swap space (KBytes) currently in use by the system. Swap space is used when memory is in short supply. Typically, entire processes are swapped out to secondary storage. Comparing this value to totalSwap (in Concord's kernelConfig group) can help indicate if additional swap space is needed.
Total system swap space in KBytes. Swap is a region on secondary storage (primarily disks) that is used for swapping and paging. Swapping is a memory management policy in which entire processes are written to secondary storage when memory is in short supply. When a process begins, all its pages are in virtual memory; only those pages actively being used are brought in from disk or swap.
| Copyright © 2013 CA. All rights reserved. |
|