Cluster performance report

OceanBase Cloud Platform (OCP) can generate performance reports for OceanBase clusters. A cluster performance report provides you with information about the cluster load, performance, and SQL statements within a specified period. The report also helps you learn about the cluster running status and troubleshoot performance-related problems as soon as possible.

Background

OCP can generate cluster performance reports. The Reports list on the Performance Reports page includes the following columns: Report Name, Snapshot Start Point, Snapshot End Point, Created By, Generation Status, Generated At, and Actions. The Actions column provides the View Report and Download Report options.

• You can sort the reports by Snapshot Start Point, Snapshot End Point, and Generated At.

• You can filter the reports by Generation Status.

Generate a performance report

1. Log on to the OCP console. In the left-side navigation pane, click Clusters.

2. Find the target cluster in the Clusters list and click the cluster name. The Overview page appears.

3. In the left-side navigation pane, click Performance Report. The Performance Report page appears.

4. Click Generate Performance Report in the upper-right corner of the page that appears.

5. In the dialog box that appears, configure the following settings:

   Parameter	Description
   Report Name	Required. The name of the report.
   Snapshot Start Point	You can select only a point in time in the past 30 days. The available points in time are fixed and cannot be customized.
   Snapshot End Point	You can select only a point in time within 24 hours after the time you specified for Snapshot Start Point. The available points in time are fixed and cannot be customized.

6. If the SQL monitoring data of a tenant is abnormal in the performance report, select the tenant from the View OceanBase Tenant ASH Report drop-down list to go to the ASH Report page of the tenant.

7. Click OK. A report is generated for the performance between the snapshot start point and the snapshot end point. You can view the report in the Reports list and download it.

View and download a report

OCP shows a list of performance reports generated in the last 90 days.

• To view a performance report, click View in the Actions column.

• To download a performance report in the .html format, click Download in the Actions column.

You can view the following information in a performance report:

• Basic information

  Parameter	Description
  Basic Cluster Information	Includes the cluster version, cluster type, number of hosts, number of tenants, and start time and end time of the last major compaction.
  Node Basic Information	Includes the host IP address, zone, availability of Root Server, status, service start time, and time of last deletion.
  Node Resources	Includes the host IP address, zone, total CPU cores, allocated CPU cores, allocated CPU resources in percentage, total memory in MB, allocated memory in MB, allocated memory in percentage, disk usage in GB, and disk usage in percentage.
  Node Partitions	Includes the host IP address, zone, associated tenants, number of partitions, and number of leader partitions. Note that the number of partitions includes the number of leader partitions.
  Tenant Resources	Includes the tenant name, tenant ID, tenant type, resource unit configuration, zone priority, locality, and list of servers.
  Tenant Capacity	Includes the associated tenants, tenant ID, associated hosts, and data capacity.

• Workload Statistics: The table includes the following columns: Metrics, Cumulative Value, Start Value, End Value, Maximum Value, Maximum Value Occurred At, Minimum Value, Minimum Value Occurred At, and Average Value.

• SQL statistics

  Parameter	Description
  SQL Sorted by Total Response Time	Top 10 SQL statements with the longest response time in the cluster. Top 10 SQL statements with the longest response time in the tenant.
  SQL Sorted by CPU Time	Top 10 SQL statements with the longest CPU time in the cluster. Top 10 SQL statements with the longest CPU time in the tenant.
  SQL Sorted by I/O Latency	Top 10 SQL statements with the longest I/O wait time in the cluster. Top 10 SQL statements with the longest I/O wait time in the tenant.
  SQL Sorted by Network Latency	Top 10 SQL statements with the longest network latencies in the cluster. Top 10 SQL statements with the longest network latencies in the tenant.
  SQL Sorted by Logical Reads	Top 10 SQL statements with the most logical reads in the cluster. Top 10 SQL statements with the most logical reads in the tenant.
  SQL Sorted by Physical Reads	Top 10 SQL statements with the most physical reads in the cluster. Top 10 SQL statements with the most physical reads in the tenant.
  SQL Sorted by Executions	Top 10 most executed SQL statements in the cluster. Top 10 most executed SQL statements in the tenant.
  SQLs Sorted by Number of Remote Plan Executions	Top 10 most executed SQL statements in the cluster. Top 10 most executed SQL statements in the tenant.
  SQLs Sorted by Distributed Plan Executions	Top 10 most executed SQL statements in the cluster. Top 10 most executed SQL statements in the tenant.
  SQL Sorted by Affected Rows	Top 10 SQL statements affected most rows in the cluster. Top 10 SQL statements affected most rows in the tenant.
  SQL Texts	Displays the information of all the preceding SQL statements.

• Lock statistics

  Parameter	Description
  Row Lock Statistics	The number of applications, waits, and timeouts, and wait time of row locks in a tenant.
  Latch Statistics	You can view the following statistics on latches in a tenant: Get: indicates the number of successful willing-to-wait latch requests. Miss: indicates the number of failed willing-to-wait latch requests. Sleep: indicates the number of failed willing-to-wait latch requests that result in process sleeps. Spin Get: indicates the total number of spins in the tenant. Immediate Get: indicates the number of successful immediate latch requests. Immediate Miss: indicates the number of failed immediate latch requests.