Indicators

Indicators are computed metrics. They provide an indication of how a system is performing from the capacity and forecasting perspective. They are derived information and not measurements. With indicators, you can get additional insight into the status of the managed entity. Indicators are used individually (For example, in capacity views) or as part of Optimizer rules to compute recommendations.

Indicators are computed internally by the Auto Forecasting Service (AFS). The Near Real Time - Warehousing (NRT-WH) service notifies the AFS of the presence of new incoming data and the AFS then updates the indicators. Indicators are computed using fresh data imported into TrueSight Capacity Optimization. Some indicators are computed for three time periods, namely last 1, 7, and 30 days. Others are computed only for last 30 days.

Some indicators are configurable. You can configure the indicator parameters and thresholds from the Indicators page in the TrueSight Capacity Optimization console (Click Administration > OPTIMIZER > Thresholds > Indicators). For more information, see Configuring and managing thresholds for metrics and indicators.

Recalculating the spare VM basis

The following table lists and describes the indicators.

Oops, it seems that you need to place a table or a macro generating a table within the Table Filter macro.

UOM: unit of measurement

All = VMware, AWS, Azure, Google Cloud, Hyper-V, KVM, Xenserver, Standalone systems

Indicator Description UOM Supported cloud provider/technology

Bottleneck Resource

(IND_BNECK_RESOURCE)

A resource that is currently limiting the system. AFS evaluates this indicator as the minimum value of available days to saturation. Possible values: CPU, MEMORY, STORAGE, or MANY (more than one resource is saturated).

-

VMware - Virtual Cluster

Composite Memory Consumption

(IND_COMPOSITE_MEMORY_CONSUMPTION)

An estimate of the amount of memory that is effectively in use in the system.

The consumed memory is too high, while the active memory gives too low a picture of usage. The "composite" memory consumption is a more realistic estimate of the amount of memory that is effectively in use in the system. It considers both Active and Consumed memory numbers reported by vSphere, and adjusts these numbers using memory pressure due to contention, as measured by swapping, ballooning, and the total memory overcommitment in the system.

This indicator is based on the Memory Pressure Index (IND_MEMORY_PRESSURE) indicator. You can configure the following thresholds for this indicator:

Ratio of consumed memory by active memory threshold
Memory overcommitment threshold

Bytes

VMware: Virtual Cluster, Virtual Host, Capacity Pool

Composite Memory Utilization %

(IND_COMPOSITE_MEMORY_UTILIZATION)

An estimate (in percentage) of the amount of memory that is effectively in use in the system. It is a percentage calculated by using the value in the Composite Memory Consumption (IND_COMPOSITE_MEMORY_CONSUMPTION) indicator over the total memory of the system.

PCT

VMware: Virtual Cluster, Virtual Host, Capacity Pool

CPU Demand

(IND_CPU_DEMAND)

An estimate of the CPU demand that is based on the number of CPU cores used by the VM in the last 30 days.

This indicator is used to determine whether the VM is overallocated for CPU.

The subresources for this indicator for all supported technologies except Azure are calculated as:

Aggressive: Resource utilization in the server is computed by considering the average value of hourly samples. Spikes in resource utilization within the hour are not considered. Then, 95^th percentile of the hourly value over the last 30 days is computed for each resource to generate the configuration of the target virtual machine or instance type. This subresource is computed on the CPU_UTIL [avg] metric.
Balanced: Resource utilization in the server is computed by considering the 90^th percentile value of hourly samples. Only 10% of spikes in utilization within the hour are not considered. Then, 95^th percentile of the hourly value over the last 30 days is computed for each resource to generate the configuration of the target virtual machine or instance type. This subresource is computed on the CPU_UTILMHZ_HM [avg] metric.
Conservative: Resource utilization in the server is computed by considering the 99^th percentile value of hourly samples. Only 1% of spikes in utilization within the hour are not considered. Then, 95^th percentile of the hourly value over the last 30 days is computed for each resource to generate the configuration of the target virtual machine or instance type. This subresource is computed on the CPU_UTILMHZ_HM [max] metric.
Balanced-Aggressive: It is generated only if the CPU utilization has a sample period of 15 minutes or less (more granular than an hour). Resource utilization in the server is computed by considering the max value of CPU Utilization of hourly samples. Then, 95^th percentile of the hourly value over the last 30 days is computed for each resource to generate the configuration of the target virtual machine or instance type.
Note: This subresource is not configurable in the optimizer rule settings. When balanced behavior is selected in the settings but detailed statistics for hourly data samples are not available for the server, the optimizer rule uses the balanced-aggressive subresource.

^*The subresources for this indicator for Azure are calculated as:

Balanced: 95^th percentile of the hourly value over the last 30 days is computed for each resource on the CPU_USED_NUM_HM [avg] metric.
Aggressive: 95^th percentile of the hourly value over the last 30 days is computed for each resource on the CPU_UTIL [avg] metric.
Conservative: 95^th percentile of the hourly value over the last 30 days is computed for each resource on the CPU_USED_NUM_HM [max] metric.
Balanced-Aggressive: 95^th percentile of the daily value over the last 30 days is computed for each resource on the CPU_USED_NUM [avg] metric.

Note: This functionality is available only when you apply Cumulative Hotfix 3 or later on Patch 2 (20.02.02) of TrueSight Capacity Optimization 20.02.

CPU

VMware - Virtual Machine
AWS - Virtual Machine
Hyper-V - Virtual Machine
KVM - Virtual Machine
Xenserver - Virtual Machine
Standalone systems
Azure - Virtual Machine^*

CPU Density

(IND_CPU_DENSITY)

The ratio of total vCPUs configured on the VM (VCPU_ON_NUM) to the total number of physical cores (CPU_NUM) in the cluster.

VCPU_ON_NUM: Calculated as the 95^th percentile over the last 30 days
CPU_NUM: Current day (today) values.

Example: For a virtual cluster with 8 cores and 36 vCPUs, CPU density = 36 / 8 = 4.5.

QTY

VMware: Virtual Cluster, Virtual Host
Hyper-V - Virtual Host

CPU Pressure Index

(IND_CPU_PRESSURE)

An index between 0 and 100 that is calculated by using CPU utilization and other CPU based metrics (such as CPU_READY, CPU_QUEUE_LENGTH, and so on). Higher the value, higher the pressure. For more information, see Computation details for VMware metrics and Computation details for AIX metrics.

You can configure the following thresholds for this indicator:

For Virtual Cluster - VMware and Virtual Host - VMware:
- CPU Ready threshold for VMs
- Minimum VMs considered for CPU Ready impact

For PowerVM Host - AIX:
- CPU Queue Length threshold for partitions
- Minimum Partitions considered for CPU Queue Length impact

QTY

VMware: Virtual Cluster, Virtual Host, Capacity Pool,
AIX PowerVM: Virtual Host, Capacity Pool
KVM: Virtual Host, Capacity Pool
Hyper-V - Virtual Host

CPU Size

(IND_CPU_SIZE)

The total size (processing capacity) of the physical CPU core configured for the VM, measured in MHz per CPU fractions.

MHz/CPU fractions

All

CPU Used

(IND_CPU_USED)

The total CPU (processing) capacity utilized by the VM, measured in MHz per CPU fractions.

MHz/CPU fractions

All

Days Idle

(IND_DAYS_IDLE)

The number of days for which a VM has been idle in the last 90 days.

A VM is considered idle if it is powered on, and if its CPU utilization, network rate, and disk rate values are less than their corresponding thresholds.

This indicator has three subresources: Idle_time, Powered_On_Time, and Powered_Off_Time. These subresources are computed based on the following metrics:

Idle_time: CPU_UTIL, NET_BIT_RATE (for VMware)/NET_BYTE_RATE (for other technologies), DISK_TRANSFER_RATE, UPTIME metrics
Powered_On_Time: UPTIME metric
Powered_Off_Time: UPTIME metric

You can configure the following thresholds for this indicator:

CPU utilization threshold
Network rate threshold
Disk I/O rate threshold

This indicator is used to generate the Idle VM recommendation.

Days

VMware - Virtual Machine
Hyper-V - Virtual Machine
AWS - Virtual Machine
Azure - Virtual Machine

Days to Saturation

(IND_DAYS_TO_SATURATION)

The estimated number of days before the CPU, memory, and storage resources achieve the configured threshold.

A certain amount of historical data is required to provide accurate Days to Saturation values. By default, the last 30 days of historical data is used for forecasting saturation and the minimum 5 days of data is required to calculate the utilization trend of resources. You can modify these settings on the indicators page in the Administration > Optimizer section.

The following values are displayed for the Days to Saturation (DTS) indicator in the Workspace and Views tabs:

Workspace	Views	Description
999,99	Not Saturated	No saturation predicted or occurred in last 30 days.
-1	Saturated	Saturation of resource has exceeded threshold in last 30 days.
<Number of days>	<Number of days>	Number of days to reach capacity shortfall or saturation.

Note: If a system is marked as saturated in the last 30 days, the alert remains active until this period is over, though the system turns not saturated during this period. For example, if the system was marked as saturated 25 days back, the system will still be displayed as saturated for the next 4 days.

Configurable parameters for DTS:

Minimum number of days needed to calculate trends - Default is 5 days.
(Displayed when Use long term data to compute indicators is enabled) Number of months of long term historical data
Based on the configuration of the long term historical data, the DTS indicator is computed differently:

- When long term history is not enabled - The last 30 days of historical data is used for forecasting saturation. The robust linear algorithm is applied for estimating the number of days to saturation.
- When long term history is enabled - The user-specified historical time period in the Optimizer thresholds configuration is used for forecasting saturation. If the system has more than six months of data, the yearly time shift algorithm is applied for estimating the number of days to saturation for the next 180 days. Else, the system falls back to the robust linear algorithm. This computation of days to saturation is applicable for all system types in all capacity views.
Ignore past saturation - Default is No.
- When ignore past saturation is enabled - All alerts related to earlier saturations are disabled and not displayed in the alerts. Only upcoming saturations are displayed in the alerts. Also, saturated systems are not listed in the Recommendations page in the view.
  Note: The Ignore past saturation option is available only when you apply Patch 2 (20.02.02) of TrueSight Capacity Optimization 20.02.

The warning thresholds are based on the following metrics:

CPU - CPU_UTIL
Memory - MEM_UTIL
Storage - TOTAL_DSTORE_USED

Click here to know the metrics that are used to compute this indicator for vSphere

The following table lists the metrics that evaluate the Days to Saturation parameter for the vSphere platform. The calculations use absolute values and not percentage metrics.

Resource ↓	Calculation ↓
CPU Used	`CPU_UTILMHZ`: Calculated as: (CPU_UTIL * CPU_TOTAL_MHZ) [CPU_UTIL * (CPU_MHZ * CPU_NUM)]
CPU Size	`CPU_TOTAL_MHZ`: Calculated as: (CPU_MHZ * CPU_NUM)
Memory Used	`MEM_CONSUMED`: Calculated as: (MEM_UTIL * TOTAL_REAL_MEM)
Memory Size	TOTAL_REAL_MEM
Disk Used	`TOTAL_DSTORE_USED`: Calculated as: (TOTAL_DSTORE_SIZE * TOTAL_DSTORE_UTIL) (TOTAL_DSTORE_SIZE - TOTAL_DSTORE_FREE) `TOTAL_FS_USED`: Calculated as: (TOTAL_FS_SIZE * TOTAL_FS_UTIL)
Disk Size	TOTAL_DSTORE_SIZE

Days

All

Efficiency Score

(IND_EFFICIENCY_SCORE)

Indicates the overall efficiency of the container pool, produced by combining the efficiency of CPU, memory, and storage resources of the container pool. Under-used resources, Idle VMs, and old snapshots are considered as the inefficient use of resources.

QTY

AIX PowerVM: Virtual Host, Capacity Pool
VMware: Virtual Cluster, Virtual Host, Capacity Pool
KVM: Virtual Host, Capacity Pool
Hyper-V: Virtual Host, Capacity Pool

Reserved Instance Eligible Hours

(IND_ELIGIBLE_HOURS_RESERVEDINSTANCE)

The total number of compute hours in a region that is considered 'enough' or that qualifies to be reserved.

The eligible hours value is calculated for all instances in a region, per instance type, per operating system.

This indicator is used to generate the AWS reserved instance recommendation.

Hours

AWS - Virtual Machine

Memory Spare

(IND_MEM_SPARE¹)

Amount of spare memory available from the total size of physical memory configured for the VM.

Computation for Windows VMs:
IND_MEM_SPARE = TOTAL_REAL_MEM – MEM_USED

Computation for Unix/Linux VMs:
TOTAL_REAL_MEM – MEM_REAL_USED

where, TOTAL_REAL_MEM data is taken from the VM and the MEM_USED and MEM_REAL_USED data is taken from the Virtual Node.

Bytes

VMware - Virtual Machine

Memory Demand

(IND_MEMORY_DEMAND)

An estimate of the memory demand that is based on memory utilization of the VM in the last 30 days. This indicator is used to determine whether the VM is overallocated for memory.

For vSphere VMs, the memory demand indicator is based on 95th percentile value of the MEM_REAL_UTIL or MEM_ACTIVE metric.
When using the MEM_ACTIVE metric, the memory demand is calculated such that the MEM_ACTIVE value is increased by a factor but never exceeds the MEM_CONSUMED[pctile95] value.

IND_MEMORY_DEMAND = Real Memory Utilization (if collection inside OS is available) OR min{Active Memory * (1 + vSphere VM Active Memory increase factor) , Consumed Memory}

You can configure the vSphere VM Active Memory increase factor threshold for this indicator.

For other VMs, it is based on 95th percentile value of inside OS memory if available, else based on consumed memory of VM.

Bytes

VMware - Virtual Machine
Hyper-V - Virtual Machine
AWS - Virtual Machine
Azure - Virtual Machine

Memory Size

(IND_MEM_SIZE)

The total size of physical memory configured for the VM, measured in BYTES.

Bytes

All

Memory Used

(IND_MEM_USED)

The total amount of physical memory utilized by the VM, measured in BYTES.

Bytes

All

Memory Pressure Index

(IND_MEMORY_PRESSURE)

An index between 0 and 100 that takes ino account memory usage and contention, and provides a realistic picture of the memory pressure of a given system.

This indicator is used to compute the Composite Memory Consumption (IND_COMPOSITE_MEMORY_CONSUMPTION) and Composite Memory Utilization % (IND_COMPOSITE_MEMORY_UTILIZATION) indicators. For more information, see Computation details for VMware metrics and Computation details for AIX metrics.

You can configure the following thresholds for this indicator:

For PowerVM Host - AIX:
- Memory page rate threshold for partitions
- Minimum Partitions considered for Memory page rate impact
For Virtual Machine - VMware:
- Minimum swap rate exceed count
For VMware Clusters, Hosts and Capacity Pools:
- Ratio of consumed memory by active memory threshold
- Memory overcommitment threshold

QTY

AIX PowerVM - Virtual Host
VMware: Virtual Cluster, Virtual Host, Virtual Machine, Capacity Pool
KVM: Virtual Host, Capacity Pool
Hyper-V - Virtual Host

Oversubscription

(IND_OVERSUBSCRIPTION)

This indicator can be classified into the following types:

IND_OVERSUBSCRIPTION[Memory]: The ratio of total memory configured on the VM to the total memory in the cluster. It is calculated as:
IND_OVERSUBSCRIPTION[MEMORY] = [SUM of TOTAL_REAL_MEM of VMs in cluster] / [TOTAL_REAL_MEM]
IND_OVERSUBSCRIPTION[Storage]: The ratio of storage provisioned for the VMs to the total shared storage in the cluster. It is calculated as:
IND_OVERSUBSCRIPTION[STORAGE] = [SUM of TOTAL_STORAGE_PROVISIONED of VMs in cluster] / [TOTAL_DSTORE_SIZE]


TOTAL_REAL_MEM, TOTAL_DSTORE_SIZE, TOTAL_STORAGE_PROVISIONED

are calculated on the current day's (today) values.

QTY

VMware: Virtual Cluster, Virtual Host

Risk Score

(IND_RISK_SCORE)

Indicates the overall risk of the container pool, produced by assigning penalty points for key capacity and performance metrics of the container pool. Metrics from CPU, memory, storage are taken into account.

QTY

All: Virtual Machine
AIX PowerVM, VMware, KVM, Hyper-V: Virtual Host, Capacity Pool
Service pools
Standalone systems

Spare VMs

(IND_SPARE_VM)

An estimate of the additional VMs that can be added to a cluster based on a typical VM configuration. Using this estimate, you can identify the clusters with residual or spare capacity.

You can configure the following thresholds for this indicator:

Use Composite Memory Utilization indicator instead of Memory Utilization metric
Default VM CPU usage
Default VM memory usage
Default VM storage usage

QTY

All: Virtual Cluster, Virtual Host
VMware - Virtual Machine
OpenStack cloud: Host Aggregate, Availability Zone, Region

Storage Demand

(IND_STORAGE_DEMAND)

An estimate of the storage demand that is based on storage utilization (TOTAL_FS_UTIL[pctile95] or TOTAL_FS_USED[pctile95] metric) of the VM in the last 30 days.

This indicator is used to determine whether the VM is overallocated for storage.

Bytes

VMware - Virtual Machine
AWS - Virtual Machine
Azure - Virtual Machine

Storage Size

(IND_STORAGE_SIZE)

The total amount of physical disk storage configured or available for the VM, measured in BYTES.

Bytes

All

Storage Pressure Index

(IND_STORAGE_PRESSURE)

An index between 0 and 100 that takes into account the storage utilization and provides a realistic picture of the storage pressure of a given system. For more information, see Computation details for VMware metrics.

You can configure the following thresholds for this indicator:

For VMware Clusters and Hosts:
- I/O latency threshold
- Minimum Datastores considered for I/O latency impact

QTY

VMware:
- Virtual Cluster
- Virtual Host
- Capacity Pool
AIX - Capacity Pool
Hyper-V - Virtual Host

Storage Used

(IND_STORAGE_USED)

The total amount of physical disk storage utilized by the VM, measured in BYTES.

Bytes

All

Usage Score

(IND_USAGE_SCORE)

Indicates the overall usage of the container pool, produced by combining the usage of CPU, memory, storage resources of the container pool.

QTY

All: Virtual Machine
AIX PowerVM, VMware, KVM, Hyper-V: Virtual Host, Capacity Pool
Service pools
Standalone systems

vCPU Size

(IND_VCPU_SIZE)

The total number of virtual CPUs configured on or for the VM. This number includes the vCPUs that may or may not be currently in use by the VM.

CPU

All

vCPU Used

(IND_VCPU_USED)

The total number of virtual CPUs currently in use by the VM.

CPU

All

Virtualization Readiness Index

(IND_VIRT_READINESS)

This indicator is deprecated.

Index (in percentage) that enables you to choose the best candidate VMs for virtualization.

It is calculated using the following formula:

VRI = (1 - min(1,AVG(CPU_UTILMHZ)/MAX_CPU_GHZ))
    * (1 - min(1,AVG(MEM_CONSUMED)/MAX_MEM_AVAILABLE))
    * (1 - min(1,AVG(DISK_IO_RATE)/MAX_DISK_IOPS))
    * (1 - min(1,AVG(NET_BIT_RATE)/MAX_NET_BANDWIDTH))

This formula calculates the average resource utilization for the candidate VM, and compares it with a set of reference values.

Default values are:

MAX_CPU_GHZ: 5 GHz
MAX_MEM_AVAILABLE: 8 GB
MAX_DISK_IOPS: 75K I/O operations per second (IOPS)
MAX_NET_BANDWIDTH: 1 Gigabit per second

Note: The VRI value that is computed based on the preceding formula and default values lies between 0 and 1. A VRI closer to 1 signifies that the VM is a good candidate for virtualization, due to low resource utilization. A VRI of 0 does not necessarily mean that the VM cannot be virtualized, but that its virtualization should be considered carefully.

You can configure the following thresholds for this indicator:

CPU consumption metric mandatory
CPU consumption threshold
Memory consumption metric mandatory
Memory consumption threshold
Disk consumption metric mandatory
Disk consumption threshold
Network consumption metric mandatory
Network consumption threshold

PCT

Standalone systems

¹ - Computed for VMs only if there is an established relationship (GM_ASSOCIATED_TO_VN) between the VM and its corresponding Virtual Node (VN). To create this relationship, rerun the Virtual Node Linker system task (Administration > ETL & SYSTEM TASKS > System tasks > Virtual Node Linker).

Indicators

Comments