Memory Usage (One check per server)

The Memory Usage check polls the selected Memory Performance Indicators every 5 seconds and averages this out over the monitored period (5 or 15 minutes). Where the average for any indicator exceeds the entered threshold figure an Alert is generated.

RAM Used

Threshold: Alert if average available MB <

RAM used is the total memory minus the available memory. Available memory is the sum of the following three internal memory lists.

  • Standby List Cached memory pages not in a process’s active working set are placed in the Standby List. If the same process needs the page, and it is in the Standby List, it is returned immediately.
  • Free List Memory pages that have not been used for some time by a process are cleared and placed on the Free list.
  • Zero List Memory pages that have been freed and initialized to zero are placed on the Zero List.

Unused pages of memory and pages of memory returned to the operating system when a process has either quit or its working set was reduced to comply with system resource requirements are placed on one of the above three lists. They will first go to the Standby List, then the Free list and eventually the Zero list.

Where a process makes a memory allocation request the memory manager will first look at the Zero List. If that list is empty it will then look at the Free List and if that list is empty it will clear memory pages in the Standby List for use by the process.

Pages per second (hard page faults only)

Threshold: Alert if average pages per sec >

Hard page faults occur where the system attempts to access a virtual memory page not currently in physical memory working set. This page is instead retrieved from the disk resulting in a delay.

Swap Usage (Page File Usage)

Threshold: Alert if average page file usage >

The memory pages used by a process not stored in physical RAM. Paging files can be resized to accommodate the additional memory requirements of the running processes. Problems may occur where the systems paging file is inadequately sized for the paging activity of the system.

Non-pageable memory

Threshold: Alert if non-paged pool >

Consists of memory allocated to processes that either exited abnormally, have not returned the memory allocation to the Operating System for reuse or memory in use by running processes. As such the Operating System cannot swap such memory from physical RAM to the pagefile as it is classed as in use.

Commit Charge

Threshold: Alert if average committed bytes >

The demand for memory and may be thought of as the number of bytes of physical (RAM) and virtual memory allocated to all running programs, applications and the Operating System a RAM page frame or page slot in the pagefile is committed to.

Thresholds

From Dashboard 2020.01.20 we updated the default threshold values for new Checks. Existing Check thresholds are unaffected.

Indicator Threshold Pre-2020.01.2 Threshold Threshold from 2020.01.2
RAM Used Alert if average available MB < 10MB 500MB
Pages per second (hard page faults only) Alert if average pages per sec > 20 /Sec 1000 /Sec
Swap Usage (Page File Usage) Alert if average page file usage > 70% 80%
Non-pageable memory Alert if non-paged pool > 100MB 500MB
Commit Charge Alert if average committed bytes > 1.5 *Physical RAM 1 *Physical RAM