Amazon Aurora DSQL observability concepts and usage with Amazon CloudWatch

This article explains Amazon Aurora DSQL's enhanced observability capabilities through Amazon CloudWatch Database Insights, focusing on time-based performance diagnostics aligned with the on-demand DPU pricing model.

• DSQL Active Session History (DASH) samples every active session per second and publishes 1-minute aggregated metrics to CloudWatch
• CloudWatch Database Insights provides intuitive visualization of database load by wait events and most active SQL statements
• DSQL wait events are minimal compared to PostgreSQL, including OnCpu, ClientRead, ClientWrite, Commit, storage reads, and constraint checks
• PromQL queries enable direct analysis of DASH data for identifying bottlenecks by wait events and query performance
• Application health is determined by comparing proportions of wait events over time rather than absolute numbers due to elastic scaling
• DSQL system diagnostics AI skill automates performance analysis with historical baseline comparisons and markdown reports

Time-based observability in DSQL directly correlates performance metrics with costs, enabling developers to optimize queries and manage expenses efficiently.