You are hereMetric / Metric
Metric
kb48: VMware memory monitoring
Most VMware statistics are collected via SNMP using the VMware Resources MIB. However, Longitude parses Memory metrics directly from /proc/vmwware/mem on the ESX LINUX host. This data is a more accurate reflection of memory usage than the memory data maintained under the Resources MIB.
kb94: Unix CPU percentages on HPUX inaccurate.
Longitude uses the iostat command to collect the CPU_IDLE_PCT statistic. On HP-UX, this command may return inaccurate values, which can result in CPU Busy percentages over 100%, and CPU Idle percentages less than 0.
kb93: Metrics are reported as "-1" on AIX
The RUNNABLE_PROCESSES and SWAP_OUT_RATE metrics are not collected for AIX, so the values are set to -1 to indicate that these metrics are not available on AIX for the Unix application.
kb89: How is memory usage calculated on Windows?
On Windows, physical memory usage is calculated based on the following properties collected using WMI from the root/cimv2 namespace:
- "TotalPhysicalMemory" property from the "Win32_ComputerSystem" object
- "AvailableBytes" property from the "Win32_PerfRawData_PerfOS_Memory" object
kb87: SQLServer Cache reports show greater than expected cache hit %
SQLServer Cache reports display the Buffer Manager:cache hit percentage, rather than the Cache Manager:cache hit percentage.
The values in the report correspond to the "Buffer cache hit ratio" counter of the BufferManager object for the SQLServer instance in Performance Monitor.
kb39: Value reported for Linux free memory is higher than expected
The Unix memory collector for Linux runs free -k to collect data. The output of this command looks like:
| total | used | free | shared | buffers | cached | |
| Mem: | 513920 | 476196 | 37724 | 0 | 4796 | 87840 |
| -/+ buffers/cache: | 383560 | 130360 | ||||
| Swap: | 1048568 | 160916 | 887652 |
