##production-ready node architecture
signs of a successful Node app:
it's made of several small processes that start instantly, maintain no state, and communicate via a language-agnostic message bus. each process logs to stdout and responds to unrecoverable errors by exiting immediately. a load-balancer queues requests, which are served in fixed time by scaling service levels. configuration controls behavior without any need to deploy new code. each deploy is archived into a perfect binary copy.
- sm pr -- sm mem footprint, light, not lots of blocking cpu ops
- st inst -- gives level of fault tolerance
- no st -- if process dies, state dies with it
- ex im -- just exit and start over when you have a prob
- l-b -- use something like ngnix or proxy
- co co be -- conf options should be part of env
- pe bi co -- archive your deploys!
- the apps also drop down to lang-indy text or JSON interfaces for doing and recording work