Tuning Management Pack for VMware

Veeam MP for VMware comes with a number of monitors whose default thresholds are configured based on best-practice and industry-standard values. However, Veeam MP for VMware allows you to tune certain monitors to better meet your monitoring policy requirements. It is recommended to review the following list of such monitors before you add connections to your VMware systems.

This list is a subset of all available Veeam MP for VMware monitors that are considered to be the most appropriate for pre-deployment tuning. For the full list of all Veeam MP for VMware performance monitors, see Performance Analysis Monitors. For more information on how to configure overrides for these advanced monitor types, see Overriding Performance Analysis Monitors section.

Monitor	Default Thresholds	Tuning
Veeam VMware: Datastore Provisioning Analysis	diskPressure > 125 (warning)	This monitor tracks disk pressure for datastores — that is, storage overallocation expressed as a percentage. Over-allocation (thin provisioning) is not directly a performance issue, so this monitor can be viewed as tracking IT policy for the storage environment. If you encounter thin provisioning in your environment with values greater than 125% overcommit, it is recommended to override the threshold for this monitor.
Veeam VMware: vSphere Cluster Memory Overcommit Analysis / Veeam VMware: vSphere Host Memory Overcommit Analysis	Memory Pressure > 150 (warning)	These monitors track memory pressure for clusters and hosts — that is, memory overcommit expressed as a percentage. This metric affects vSphere features such as TPS (transparent page sharing), so it does indicate when true memory overcommit is occurring. Memory overcommit does not indicate a performance issue in general. However, it does place more stress on vSphere memory management techniques. That is why these monitors can be considered as a tool tracking an IT policy for memory allocation in vSphere hosts and clusters. If you encounter memory overcommit in your environment with values greater than 150%, it is recommended to override the threshold for these monitors.
Veeam VMware: vSphere Cluster CPU Overcommit Analysis / Veeam VMware: vSphere Host CPU Overcommit Analysis	CPU Pressure > 150 (warning)	These monitors track CPU pressure for clusters and hosts — that is, CPU overcommit expressed as a percentage. This metric is based on the ratio of running virtual machine vCPUs divided by host physical CPU cores. The default value of 10 vCPU per host core is used to calculate this ratio as a percentage (100%). This default value is taken from the range between the industry standard value (of 4–8 vCPU per host core), and the vSphere maximum value (of 25 vCPU per host core) as the most common. The default value can be modified in the Veeam UI. Such modification will affect the calculation of cpuPressure metric for a Collector. CPU overcommit does not indicate a performance issue in general. However, it does place more stress on the vSphere host CPU scheduler. The real-world achievable ratio often depends on the workloads running in VMs. For example, VDI environments may achieve a much higher ratio and therefore a much higher cpuPressure rating for a host or cluster. If you encounter CPU overcommit in your environment with values greater than 150%, it is recommended to override the threshold for these monitors.
Veeam VMware: Datastore Unknown Files Analysis	unknownFilesGB > 50, warning	This monitor tracks unknown files on a vSphere datastore — the files which are unknown to the vCenter Server or not managed by the vCenter Server. These could be orphaned VM disks or snapshots, or manually uploaded files such as ISOs. If your environment contains datastores that have a specific purpose to store files not managed by the vCenter Server, such as ISO storage, it is recommended to override the threshold for this monitor.
Virtual Machine Disk Space Analysis	freeMB < 100	This monitor tracks free disk space in MB on partitions inside virtual machines. This is a multi-instance monitor that tracks all configured OS partitions even those targeted at a single VM object. Disk partitions can be included and excluded by a name in overrides for this monitor. For example, in most Linux machines the partition (boot) has less than 100MB of free space. In this case, an alert is generated immediately after monitoring starts. That is why the partition (boot) is excluded by default. It is recommended to override the threshold for this monitor according to the usage of VMs and specific disk partitions in your environment. Note: To allow Veeam MP for VMware to collect the data agentlessly, VMware Tools must be installed in the virtualized OS.
Virtual Machine Snapshot Usage Analysis	snapshotAge > 48 (hours) snapshotsSize > 2048 (MB)	This monitor tracks snapshot usage for virtual machines. Both age (in hours of the oldest snapshot) and size (total value in MB for all snapshots) are monitored. Snapshots usage depends on the needs of a system. For example, in a production environment, snapshots are usually used only on a strictly controlled or temporary basis (for example, by image-based backup solutions). On the contrary, snapshots can be widely used in development or QA environments. It is recommended to override the threshold for this monitor taking into account the specifics of your virtual environment.

Monitor

Default Thresholds

Tuning

Veeam VMware: Datastore Provisioning Analysis

diskPressure > 125 (warning)

This monitor tracks disk pressure for datastores — that is, storage overallocation expressed as a percentage. Over-allocation (thin provisioning) is not directly a performance issue, so this monitor can be viewed as tracking IT policy for the storage environment.

If you encounter thin provisioning in your environment with values greater than 125% overcommit, it is recommended to override the threshold for this monitor.

Veeam VMware: vSphere Cluster Memory Overcommit Analysis /

Veeam VMware: vSphere Host Memory Overcommit Analysis

Memory Pressure > 150 (warning)

These monitors track memory pressure for clusters and hosts — that is, memory overcommit expressed as a percentage. This metric affects vSphere features such as TPS (transparent page sharing), so it does indicate when true memory overcommit is occurring.

Memory overcommit does not indicate a performance issue in general. However, it does place more stress on vSphere memory management techniques. That is why these monitors can be considered as a tool tracking an IT policy for memory allocation in vSphere hosts and clusters.

If you encounter memory overcommit in your environment with values greater than 150%, it is recommended to override the threshold for these monitors.

Veeam VMware: vSphere Cluster CPU Overcommit Analysis /

Veeam VMware: vSphere Host CPU Overcommit Analysis

CPU Pressure > 150 (warning)

These monitors track CPU pressure for clusters and hosts — that is, CPU overcommit expressed as a percentage. This metric is based on the ratio of running virtual machine vCPUs divided by host physical CPU cores.

The default value of 10 vCPU per host core is used to calculate this ratio as a percentage (100%). This default value is taken from the range between the industry standard value (of 4–8 vCPU per host core), and the vSphere maximum value (of 25 vCPU per host core) as the most common. The default value can be modified in the Veeam UI. Such modification will affect the calculation of cpuPressure metric for a Collector.

CPU overcommit does not indicate a performance issue in general. However, it does place more stress on the vSphere host CPU scheduler. The real-world achievable ratio often depends on the workloads running in VMs. For example, VDI environments may achieve a much higher ratio and therefore a much higher cpuPressure rating for a host or cluster.

If you encounter CPU overcommit in your environment with values greater than 150%, it is recommended to override the threshold for these monitors.

Veeam VMware: Datastore Unknown Files Analysis

unknownFilesGB > 50, warning

This monitor tracks unknown files on a vSphere datastore — the files which are unknown to the vCenter Server or not managed by the vCenter Server. These could be orphaned VM disks or snapshots, or manually uploaded files such as ISOs.

If your environment contains datastores that have a specific purpose to store files not managed by the vCenter Server, such as ISO storage, it is recommended to override the threshold for this monitor.

Virtual Machine Disk Space Analysis

freeMB < 100

This monitor tracks free disk space in MB on partitions inside virtual machines.

This is a multi-instance monitor that tracks all configured OS partitions even those targeted at a single VM object.

Disk partitions can be included and excluded by a name in overrides for this monitor. For example, in most Linux machines the partition (boot) has less than 100MB of free space. In this case, an alert is generated immediately after monitoring starts. That is why the partition (boot) is excluded by default.

It is recommended to override the threshold for this monitor according to the usage of VMs and specific disk partitions in your environment.

Note: To allow Veeam MP for VMware to collect the data agentlessly, VMware Tools must be installed in the virtualized OS.

Virtual Machine Snapshot Usage Analysis

snapshotAge > 48 (hours)

snapshotsSize > 2048 (MB)

This monitor tracks snapshot usage for virtual machines. Both age (in hours of the oldest snapshot) and size (total value in MB for all snapshots) are monitored.

Snapshots usage depends on the needs of a system. For example, in a production environment, snapshots are usually used only on a strictly controlled or temporary basis (for example, by image-based backup solutions). On the contrary, snapshots can be widely used in development or QA environments.

It is recommended to override the threshold for this monitor taking into account the specifics of your virtual environment.