When an offline integration task exhibits slow performance, performance diagnosis can be utilized to examine the task's read and write durations, operations that are time-consuming, and memory garbage collection (GC) periods. This aids in identifying the causes of the task's sluggish execution.
Performance diagnosis entry
On the Dataphin home page, navigate to the top menu bar and select Development > Data Integration.
At the top menu bar of the Integration page, choose your Project and Environment (selecting an environment is required in the Dev-Prod pattern).
In the left-side navigation pane, click on Integration > Batch Pipeline. Then, in the offline integration list, select the desired offline integration task.
At the top of the editing area for the selected offline integration task, click Run or Preview.
Click View Console at the bottom of the page. In the console, select the
icon to access the performance diagnosis results.NoteIf you remain on the Console > Log Information > Raw Logs page for more than one minute, the system will prompt you to click here to view performance diagnosis. You can then click the indicated icon to proceed to performance diagnosis.
Performance diagnosis introduction
Performance diagnosis is divided into three parts: Read/Write Diagnosis, Time-consuming Operation Diagnosis, and Memory GC Time Diagnosis. Only metrics that are abnormal will be displayed. If a section does not detect any abnormalities, it will appear empty. You can review the diagnostic findings for each item in the list of abnormal metrics and click Actions in the relevant column to View Suggestions. Implement the system's recommendations to enhance the performance of the integration task.
Section | Description |
Read/Write Diagnosis | Analyzes the ratio of read (wait) and write (wait) times for the data source to determine if there are performance bottlenecks in read and write operations. Ratios exceeding 60% suggest potential read/write performance issues in the integration task. |
Time-consuming Operation Diagnosis | Identifies operations related to the data source that take an excessive amount of time, indicating performance issues. Operations taking longer than 180 seconds are flagged for potential performance concerns. |
Memory GC Time Diagnosis | Evaluates the JVM GC time ratio to assess if the memory configuration is adequate. A GC time ratio above 30% of the runtime indicates possible memory configuration issues. |