Skip to main content

AWS RDS Best Practices

reference: https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/CHAP_BestPractices.html

https://drive.google.com/file/d/1ntwmblVekoE2FUrEAJaFlolHEJydR9n4/view?usp=drive_link

Basic operational guidelines

  • use metrics to monitor memory, CPU, replica log, storage usage. (AWS CloudWatch)
  • scale up resource when approaching limits, have some buffer to accommodate sudden increases in demand
  • enable period backups, schedule for daily low in write IOPS
  • do not cache DNS for more than 30 seconds because the IP of a DB instance can change after a failover
  • do failover test
  • keep DB versions up to date
    • test in staging env
    • use Amazon RDS managed upgrades
    • schedule major version upgrades
  • use AWS suite of drivers for application connectivity
  • use enhanced monitoring to identify OS issues
  • use metrics to identify performance issues
  • establish a baseline performance metrics for "normal" workload
  • establish a baseline for "replica lag" if using multi-AZ DB
  • create a performance analysis report

RAM recommendations

  • RAM should be good enough so that working set should reside almost completely in memory
  • use CloudWatch to monitor ReadIOPS, it should be stable overtime

Best practices for working with PostgreSQL

  • Loading data into a DB instance
  • PostgreSQL autovacuum feature
Back to top