AI_SERVER

AI_SERVER is an independent from the anomaly_detection feature. In addition to the data collection function of the anomaly_detection feature, the collection type, collection item, and data storage mode are added to the AI_SERVER feature. The AI_SERVER feature is used only for data collection and will be integrated into the anomaly_detection feature in the future. This feature includes the server and agent components. The agent must be deployed on the database node for data collection, while the server is deployed on an independent node for data collection and storage.

The data storage modes include sqlite, mongodb, and influxdb.

Table 1 describes the collection items.

Table 1 Collection items

Collection Type	Collection Item	Description
database	work_mem	GUC parameter related to the database memory. This parameter is used to check whether the allocated space is sufficient for SQL statements involving sorting tasks.
	shared_buffers	GUC parameter related to the database memory. Improper setting of shared_buffer will deteriorate the database performance.
	max_connections	Maximum number of database connections.
	current connections	Number of current database connections.
	qps	Database performance metrics:
os	cpu usage	CPU usage.
	memory usage	Memory usage.
	io wait	I/O wait event.
	io write	Data disk write throughput.
	io read	Data disk read throughput.
	disk used	Size of the used disk space.

For details about the deployment mode, see AI_MANAGER.