eG Monitoring
 

Measures reported by RHEVHostMemTest

A contention for memory resources on the RHEV hypervisor can affect the memory allocation to VMs, which in turn can adversely impact the performance of the applications operating on the VMs. It is therefore imperative that you closely observe how the hypervisor uses the physical memory available to it, so that you can proactively determine a potential memory crunch. This can be achieved with the help of the RHEVHostMemTest. This test periodically monitors the memory usage of the hypervisor, checks whether adequate free memory is available to the hypervisor, and if not, promptly alerts users to the same. In the process, the test also reveals the top memory VMs executing on the hypervisor, checks swap memory usage, and also reports whether/not memory has been overcommitted by the hypervisor. This way, the test also points you to the probable reasons for the memory erosion (if any) - is it owing to memory-starved VMs on the hypervisor? or is it because the hypervisor has overcommitted memory?

The measures made by this test are as follows:

Measurement Description Measurement Unit Interpretation
Tot_mem Indicates the total physical memory available in this RHEV hypervisor. GB  
Free_mem Indicates the amount of physical memory currently unused on the hypervisor. GB Ideally, the value of this measure should be high. A low value or a consistent decrease in this value could indicate a gradual memory erosion, which can consequently affect the host as well as VM performance.
Used_mem Indicates the amount of physical memory currently being utilized by the hypervisor. GB Ideally, the value of this measure should be low. A very high value or a value that grows dangerously close to the Total physical memory of the host is a cause for concern.
Overhead_mem Indicates the amount of memory overhead on this RHEV hypervisor. GB The value of this measure is typically the difference between the Total physical memory and the sum of the Used physical memory and the Free physical memory measures.
Used_mem_pct Indicates the percentage of total physical memory that is being used by the hypervisor. % Ideally, the value of this measure should be low. A high value or a value close to 100% indicates a contention for memory resources on the hypervisor. Use the detailed diagnosis of this measure to know the memory configuration of each VM on the RHEV hypervisor, and how every VM is using the configured memory. Memory-hungry VMs can thus be isolated.
Free_mem_pct Indicates the percentage of percentage of total physical memory that is currently free on the hypervisor. % Ideally, the value of this measure should be high. A low value or a consistent decrease in this value indicates a contention for memory resources on the hypervisor.
Buff_mem Indicates the total I/O memory buffers in the hypervisor. MB  
Cache_mem Indicates the total OS cached memory in the RHEV hypervisor. MB  
Tot_swap_mem Indicates the total amount of swap memory on the hypervisor. GB  
Used_swap_mem Indicates the amount of swap memory that is currently utilized by the hypervisor. GB Significant or consistent memory swapping indicates that the hypervisor is severely overcommitted and that performance degradation is imminent or actively occurring.
Free_swap_mem Indicates the amount of swap memory that is currently unused on the hypervisor. GB A high value is desired for this measure, as a consistent decrease in this value is a sign of excessive swap usage, which in turn signals a memory bottleneck.
Chache_swap_mem Indicates the total swap memory that the hypervisor has cached. GB  
Used_swap_pct Indicates the percentage of swap memory used by the hypervisor. % A high value is indicative of a contention for memory resources on the hypervisor.
Is_overcommit Indicates whether the memory of this RHEV hypervisor is over committed or not.   Hypervisor memory is over-committed when the total memory space allocated (memory granted) to powered-on VMs, plus hypervisor memory overhead, is greater than the amount of total physical memory available to the host. A severe memory overcommitment can cause serious performance degradations.

If the value of this measure is Yes, it indicates that the hypervisor memory is overcommitted. The value No indicates that there is no overcommitment of memory.

The numeric values that correspond to these measure values are indicated in the table below:

Numeric Value Measure Value
1 Yes
0 No

Note:

By default, this measure reports one of the Measure Values listed in the table above. The graph of this measure however will represent the memory overcommitment status using the numeric equivalents - '0' or '1'.