Vendor-Agnostic Cloud-Native Apps
summary
Startup story shows DynamoDB → 5–8 ms p99, 70 ms API overall and a migration blocker when trying to move to GCP despite 20M monthly users, ~1k rps. Concrete recommendations: prefer Postgres in containers, JDBC/SMTP/OTel, Terraform or Pulumi, Earthly/Dagger for portable CI, avoid Data API and SDKs that bake secrets manager or X-Ray into app code, write small adapters (put/get) for Kinesis/Kafka, run full stack locally with same service versions as a litmus test.
AI-written summary · about
chapters (9)
- 0:00Intro
- 1:15How Lock-in Creeps in
- 3:30Why Vendor Agnostic Solutions
- 4:30Conscious Decision vs Accidental Lock-in
- 7:49Strategies for Designing Vendor Agnostic Applications
- 11:16Startup Story: Careful this time
- 15:33Abstract out vendor specific code
- 17:31The Litmus Test
- 19:35Summary Choosing the right stack
related talks
58mFaster, Cheaper, Greener: Pick Three
30mBuilding Applications in the Cloud: Best Practices
35mConvergence of Serverless APIs & Compute
16mBuild for real time on AWS: Thrive in an unpredictable market
27mScalable by Design: How to Build an Elastic Cloud Infrastructure...
27mDon’t put all of your APIs in one cloud
01h 02mDiamond Sponsor Hall
06mAccelerate Application Delivery with a Cloud-Native Mindset