Kubernetes Cost and Workload Profiler

The Efficiency and Risk Profiler analyzes workloads running on a Kubernetes cluster to identify performance and reliability risks and inefficient configurations. It analyzes various workload configuration settings and operations metrics, including the pod's quality of service (QoS) and replica count, as well as resource allocation, limits, and actual usage.

Supported Environments

AWS is the only supported environment for the Kubernetes Cost and Workload Profiler.

Efficiency and Risk Profile View

To view the Efficiency and Risk Profile view:

From Observe, navigate to Kubernetes > Workloads.
From the Workloads list view, click the Name of a workload where Workload Type is Deployment and displays metrics (not blank values).
Scroll down to the Application Resource Optimizer widget.
Select the Profiler tab.
Click > Show more to view the following workload data:

Analysis Outcomes

While deployment analysis outcomes always include the Analysis conclusion, Reliability Risk, and Efficiency Rate, the specific Opportunities and Cautions are included only when enough data is available to decide. Findings and ratings are based on the assessment of the data available for the workload during the analysis period.

Heading	Description
Analysis	The most pertinent of all findings drawn.
Reliability Risk	Level of risk to workload performance and reliability. Risk rankings are low, medium, high, and critical.
Efficiency Rate	Workload efficiency is calculated as a percentage relative to allocated resources.
Opportunities*	Analysis findings of potential targets to improve the workload.
Cautions*	Highlights likely causes of problems identified during analysis.
Average Replicas	The number of pods (replicas) running for this workload averaged over the analysis period.
CPU Utilization	The percentage of CPU that was requested averaged over the analysis period. Note that it may exceed 100% when the pod uses more than the requested value.
Memory Utilization	The percentage of requested memory that was used was averaged over the analysis period. Note that it may exceed 100% when the pod uses more than the requested value.

*Displays only when an applicable finding is discovered.

Opportunities

Opportunities Rating	General Description
Improve performance/reliability	Workload performance and reliability can be improved by adjusting its resources.
Improve efficiency by up to X%	Adjusting resources can improve workload efficiency. Profiler estimates impact potential (for example, a percentage or multiplier) to include in the opportunity statement when possible.

Cautions

Cautions Rating	Description
Pod QoS class is X	Indicates that the pod's quality of service may cause reliability or performance risks. Using pod classes below Guaranteed makes it more likely that the pod will not get sufficient resources to operate or can be evicted to make room for higher priority QoS classes.
Resource utilization significantly exceeds allocation	The profiler detected a memory or CPU utilization rate that may be causing performance or reliability issues. Increasing allocated resources and having a high QoS class may help resolve these issues.
Resource utilization exceeds the allocation
Resource utilization close to allocation
Idle application	Usage rate indicates that an application can be shut down. Applications with low or no load cannot be fully analyzed.
Low request rate
Fewer than 2 replicas	The application lacks sufficient redundancy to provide services in case of hardware or software failure. Increasing the number of pods may improve reliability.