ApsaraDB for POLARDB provides the performance insight feature, which focuses on monitoring the load, analyzing the load, and optimizing the performance of an ApsaraDB for POLARDB cluster. The feature helps you easily evaluate database loads, find the causes for performance problems, and enhance database stability.

Scenarios

Performance insight can be applied in the following scenarios:

  • Analyze the cluster metrics

    Performance insight helps you monitor the key metrics of an ApsaraDB for POLARDB cluster. It also allows you to check the status and trend of the loads for the cluster. You can identify the sources that generate loads and the distribution of loads within a certain period from the trend charts of key metrics.

  • Evaluate database loads

    ApsaraDB for POLARDB provides the trend chart of average active sessions (AAS), which alleviates the need to analyze the complicated trend charts of various metrics. AAS trend chart shows the information of all key metrics to help you evaluate the sources that generate loads and cause performance bottlenecks. You can determine the causes for performance bottlenecks, such as high CPU usage, lock-waiting, and I/O latency, and find the corresponding SQL statement that incurs the problem.

    Note AAS is the number of average active sessions of an ApsaraDB for POLARDB cluster within a certain period. The trends of AAS reflect the changes of the loads for the cluster. In the performance insight feature, AAS is a key metric used to measure the loads for an ApsaraDB for POLARDB cluster.
  • Find the sources that cause performance problems

    You can analyze the trend chart of AAS and load sources in multiple dimensions to determine whether a performance problem is caused by improper cluster configurations or the database architecture. You can also find the corresponding SQL statement that incurs the performance problem.

Procedure

  1. Log on to the ApsaraDB for POLARDB console.
  2. Select a region.
  3. Find the target cluster and click the cluster ID in the Cluster Name column.
  4. In the left-side navigation pane, choose Diagnostics and Optimization > Performance Insight.
  5. Select filtering conditions.

Description of the metrics page

  • Trend charts of key metrics

    You can use the trend charts of key metrics to check the load status and resource bottlenecks of an ApsaraDB for POLARDB cluster.

    You can select a given time period or specify a custom time period to retrieve the trend charts of key metrics within the corresponding time period.

  • Trend chart of AAS

    After you use the trend charts of key metrics to check the load status, you can identify the load sources.

    Note max Vcores indicates the maximum number of CPU cores that can be used by an ApsaraDB for POLARDB cluster. The value determines the processing capacity of CPUs in the cluster.

    From the real-time trend chart of AAS, you can find the load sources, the time when loads occur, and the trend of loads over a period of time.

  • Load sources from multiple dimensions

    You can learn the trend of the loads for an ApsaraDB for POLARDB cluster by analyzing the trend chart of AAS. You can find the specific SQL statements that cause performance bottlenecks, and the related users, hosts, and databases.

    As shown in the lower section of the preceding figure, you can find the SQL statements that affect the loads, and the usage ratio of each statement in a specified AAS.

    Performance insight supports six dimensions of AAS. You can switch dimensions by using the drop-down list of AAS Type in the upper-right corner of the AAS page.

    Type Description
    SQL The trends of top 10 SQL statements in your business.
    Waits The trends of wait events within the specified time period.
    Users The trends of logon users.
    Hosts The trends of hostnames or IP addresses of clients.
    Databases The trends of the databases where your businesses are located.
    Status The trends of active sessions within the specified time period.