Notes from discussion: Nodejs at scale


  1. A proper way to do logging is important when multiple services are involved.
  2. Service management — how is that done?
  3. Organize mircorservices into 3 layers. One sits over data, the middleware one is one that takes data and does things, etc.
  4. Graphql is getting points because it eliminates data — microservies with a lot of orchestration layers do send not useful data to clients.
  5. Microservices can become something that hamper developer velocity. Mini repose, closely linked, contract between them — can hamper ..vel.. How? Lots of versioning, etc ~~~ this is about services managements?
  6. We can use eventbus based approach to make highly coupled services simpler. The versioning can still be a problem .. not sure how
  7. If micro service is not matcher and has a churn — it will create bad impact. Esp. when micro services share database…