Skip to main content
Cloud Insights

Insights: High-Consuming Workloads

Contributors netapp-alavoie

High-impact workloads can reduce the performance of other workloads in a shared resource. Cloud Insights provides tools to help you investigate resource saturation and impact in your environment.

Terminology

When talking about workload or resource impact, the following definitions are useful.

An Impactful or Greedy resource is one that causes a negative impact on another resource. For example, a volume experiencing very high IOPS may cause increased latency in other volumes (i.e. Impacted or Degraded resources). Impactful and Impacted resources are “peers” that utilize the same shared resource, for example, a storage pool or Volume.

A High-Consuming Workload is a workload that is currently identified as impacting other resources in the shared storage pool. These workloads drive higher IOPS, reducing IOPS in the Impacted Workloads. In Cloud Insights these are identified as Demanding Workloads.

An Impacted Workload is a workload affected by a high-consuming workload in the shared Storage Pool. These workloads are experiencing reduced IOPS and/or higher latency, caused by the Demanding Workloads.

Shared Resource Saturation is the ratio of impacting IOPS to baseline.

Baseline is defined as the maximum reported data point for each workload in the hour immediately preceding the detected saturation.

A Contention occurs when IOPS are determined to be affecting other resources or workloads in the shared storage pool.

High-Consuming Workloads

To start looking into Impactful and impacted workloads in your shared resources, click on Dashboards > High Consuming Workloads in the Cloud Insights menu.

High Consuming Workloads menu

Cloud Insights displays a list of any workloads where a saturation has been detected. Click on a workload to open the workload dashboard.

Workloads List

Click on a workload to view the details page for it. The top chart shows the activity on the shared resource (for example, a storage pool) on which the contention is occurring.

Chart showing activity on the shared resource

Below that are two charts showing the Demanding workloads and the workloads that are impacted by those demanding workloads.

Demanding workload chart
Impacted workload chart

What do I do to resolve saturation?

There are a number of steps you can take to reduce or eliminate the chance of saturation in your environment. Here are a few things you can try.

  • Move high-IOPS consumers

    Move the "greedy" workloads to less-saturated Storage Pools. It is recommended to assess the tier and capacity of these pools before moving the workloads, to avoid unnecessary costs or additional contentions.

  • Implement a quality of service (QoS) policy

    Implementing a QoS policy per workload to ensure enough free resources available will alleviate saturation on the Storage Pool. This is a long-term solution.

  • Add additional resources

    If the shared resource (for example, Storage Pool) has reached the IOPS saturation point, adding more or faster disks to the pool will ensure enough free resources available to alleviate saturation.