RDS is Crap
summary
RDS charges per-IO billing (about $0.20 per million IOs) and an IO‑optimized tier that raises instance and storage costs ~25%, making monthly bills unpredictable; extended support doubles version‑end charges so a 32‑vCPU multi‑AZ cluster goes from ~$63k/year to ~$119k then ~$175k, roughly $4M over three years if left untouched; stopped instances auto‑restart after 7 days; RDS Proxy pins connections on prepared statements; Aurora DSQL is promising but uses opaque DPU-based pricing.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 1:52Route 53 as AWS’s only true database
- 2:15What RDS includes
- 4:02AWS recommender insults users
- 8:43Unstable RDS console recs
- 9:24Why RDS is bad
- 10:28Aurora pricing flaws
- 11:54RDS ignores stop requests
- 13:19Extended support fees
- 15:48Extended support delays
- 17:32Forced upgrades
- 19:19Aurora lock-in costs
- 20:44Parameter group jail
- 21:48RDS custom value?
- 22:31Erosion of ops skills
- 24:14AWS support failures
- 25:38RDS proxy failures
- 27:01Aurora Data API limits
- 27:41Aurora Serverless pricing
- 29:07Paths away from RDS
- 31:12Staying on RDS
- 31:33Route 53 anti-RDS
- 32:38Q&A
related talks
01h 05mDatabase management made simple - Amazon RDS (Part 1 and 2)
23mNew Amazon RDS innovations in 2022
43mKeep calm and dive deep into newest database innovations
29mRunning MySQL on AWS
43mDeep dive into Amazon Aurora and its innovations
37mNavigating Managed Databases for Over a Decade as a Non-DBA
40mPushing past PostgreSQL’s vertical scaling limits - Running Amazon RDS
40mCloud Databases: Pros and Cons