Using the Workload Manager User Interface

Describes a few frequently used interface elements of Workload Manager that help you identify and troubleshoot your workload issues.

The following examples describe a few interface elements of Workload Manager that enable you to quickly identify workload problems, health issues, resource contentions, and abnormal or degraded performance problems.

Identify Workload problems and Health Issues

Quickly locate what engines are running on your clusters and what jobs and queries are failing the health tests with the following chart widgets:
  • The Usage Analysis chart widget, which displays what engines are running on the cluster, how many jobs or queries are processed by the engine, and how many jobs or queries have failed and missed their SLA.

    Location: This chart widget is found on the Cluster Summary page, which is accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting Summary.

  • The Suboptimal chart widget, which displays the distribution of jobs and the jobs and queries that failed. This chart widget enables you to visually see at a glance what issues are currently impacting your jobs or queries and how they are executing on your cluster:


    Location: This chart widget is found on the engine's page after selecting the workload cluster for analysis and then from the navigation side-bar, selecting an engine.

Identify and Address Resource Contentions

Workload Manager provides the following chart widgets that help you analyze and identify resource consumption and contention problems:
  • Location: The following chart widgets are found on the Cluster Summary page, which are accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting Summary:
    • The Resource Consumption By Services chart widget displays the CPU and memory consumption for each service across the time range you selected. Hover your mouse over the time line, to display the amount of CPU or memory, as a percentage, that is consumed by each of the cluster's services.


    • The Resource Consumption By Nodes chart widget displays the CPU and memory consumption for each node in the cluster. Hover your mouse over the time line, to display the amount of CPU or memory, as a percentage, that is consumed by each node and its services.


  • Location: The following chart widgets are found on the engine's page, which is accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting Impala:
    • The Memory Utilization chart widget, displays the aggregated maximum amount of memory that is used by the queries on any node performing the processing. It helps you identify inefficient queries that are consuming the most amount of memory during processing and if you need to allocate more memory to continue running your queries.
    • The Resource Consumption chart widget, displays the concurrent use of CPU and memory consumption for a workload across the time line you selected.


Identify and Address Abnormal or Degraded Performance Problems

Workload Manager enables you to identify and address abnormal or degraded performance problems by establishing baselines from health issues that also enable a performance comparison of your workloads. The Workload Manager baseline metrics measure the current performance of a job against the average performance of previous runs. They use performance data from 30 of the most recent runs of a job and require a minimum of three runs. The baseline comparisons start with the fourth run of a job.

The following images show some of the baseline and comparison metrics that are provided:
  • This image shows a few of the performance metrics listed in the Baseline tab:


  • This image shows the comparison between the baseline performance metrics and the current job run:


Location: The baseline for a job or a query is found on the Baseline page, which is accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting Summary. In the Trend chart widget, select the engine tab for analysis and then, depending on the engine, click Total Jobs or Total Queries. From the Jobs or Queries page, select the job or query of interest and then select the Baseline tab.

Identify Performance Trends

You can identify trends as well as baselines by analyzing your engine’s or cluster’s performance trends from the Trends chart widget and the Trend tab. Where:
  • The Trends time-series chart widget, displays more detailed metrics about the processed jobs and queries and enables you to view historical trends for analysis when you select a predefined or custom time period from the Range filter list.


    Location: This chart widget is found on the engine's page, which is accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting the engine of interest.

  • The Trends tab, displays the job or query's instances executed during the selected time range. Depending on the engine, the Trends page displays a job's historical trend from Duration, Data Input, and Data Output histogram charts or lists the runs of the query to show how its performance changes overtime.

    Location: This chart widget is found on the Jobs or Query's page, which is accessed by selecting the workload cluster for analysis and then from the navigation side-bar, selecting the engine of interest. Depending on the engine chosen, from the Trends chart widget, click Total Jobs or Total Queries and then from the Jobs or Queries page, select the job or query of interest and then select the Trends tab.