A Lambda function connects to RDS for each short operation, and concurrent invocations cause too many connections errors.
The team wants to reduce database connection pressure and failover impact without changing the SQL work in the application.
Which action is most appropriate?
Route the application through RDS Proxy and configure connection pooling limits such as MaxConnectionsPercent to protect the database from connection bursts.
Detailed explanation
Unlimited concurrency creates even more database connections and worsens the connection-limit problem.
Unlimited concurrency creates even more database connections and worsens the connection-limit problem.
DNS TTL controls name-resolution caching; it does not manage connection pooling or database connection counts.
DNS TTL controls name-resolution caching; it does not manage connection pooling or database connection counts.
RDS Proxy pools and multiplexes client connections and can limit connections to the database to protect it from oversubscription.
RDS Proxy pools and multiplexes client connections and can limit connections to the database to protect it from oversubscription.
Log storage helps investigation but does not reuse application-to-database connections.
Log storage helps investigation but does not reuse application-to-database connections.
Check official documentation
Compare the following points with the official documentation.
Amazon RDS Proxyの接続プール、接続多重化、MaxConnectionsPercentの公式説明を確認する。
Official references
What to confirm
アプリケーション接続とDB接続を分離し、接続プールで急増を吸収する構成を説明できる。
Key points
- RDS Proxy
- Connection pooling
- MaxConnectionsPercent
- ConnectionAttempts
Notes
- Environment: AWS公式SOA-C03試験ガイドとAWS公式ドキュメントの確認
- Confirm current behavior in the official SOA-C03 guide and AWS service documentation.
Foundation review
Connection boundary
RDS Proxy can accept many client connections while keeping a controlled number of connections to the database.
Overload protection
Connection limits can queue or reject excess requests instead of allowing a burst to overwhelm the database.