Why Website Login Is Slow After Migrating to RDS for PostgreSQL?

After migrating workloads to RDS for PostgreSQL, slow website logins are usually caused by the following factors:

• Increased network latency: The network path between the application and the database becomes longer (such as cross-AZ, cross-region, or public network access), increasing the round-trip time for multiple requests during the login process.
• Improper connection pool configuration: Parameters such as initial connections, maximum connections, and idle timeouts may be misconfigured, causing frequent connection creation or queueing, which lengthens the login path.
• DNS resolution differences: When connecting through a domain name, DNS resolution may take longer or resolve to a non-optimal node, affecting the speed of the first connection.
• Instance specifications and environment differences: The RDS for PostgreSQL instance may differ from the local database in specifications and parameters. Connection management components may not be enabled, or security group/firewall rules may not allow traffic, causing connection retries.

The following two actions are recommended:

• Use the RDS for PostgreSQL console to check instance performance metrics, such as CPU usage, memory usage, and connection-related metrics, to rule out resource bottlenecks.
• Because slow website logins are often related to the application, run the following command to check the current database connection status and compare differences between the local database and the RDS for PostgreSQL instance:

  SELECT * FROM pg_stat_activity;