Some Funny Things Happened on The Way to A Service Ecosystem
summary
A team builds a rack AMQP bridge and an AMQParty gem to run Rails over AMQP but the AMQP Auth service incurs extra network hops, unacceptable latency, and is abandoned. A small bulk-pricing service for Bulq.com ships successfully using a data‑science model instead of extracting monolith data. Lessons: avoid ActiveRecord callbacks, never share a DB, choose an origin service and broadcast updates, prefer HTTP for sync calls, adopt infra conventions (Logstash Nagios Terraform Chef) and have developers operate their services.
AI-written summary · about
chapters (18)
- 0:00Start
- 2:01Authentication Authorization
- 2:22Auth’s Easy, Right? Right?
- 5:15Callbacks Held Us Back
- 7:12Data Sharing Is Hard
- 7:33Share a Database?
- 11:52Don't Share Data; Broadcast It
- 12:05Product Photo Processing
- 15:22Adding Is Easier Than Extracting
- 21:39Metrics Secrets Service Discovery Health Checks Load Balancing
- 23:54Metrics? Secrets? Service Discovery? Health Checks? Load Balancing?
- 25:02Rails Application Template
- 26:31Scrap Your Callbacks
- 27:43Pick Something With Impact
- 27:57Use Boring Tech
- 28:18To Share Data, Broadcast It
- 28:46Operate What You Deploy
- 29:39Questions - Hallway Track Chris Hoffman
related talks
40mService Architectures for Mere Mortals
27mScaling a Rails Monolith w/Event Driven Microservices
22mRack-AMQP: Ditch HTTP Inside SOA!
27mRuby Messaging Patterns
33mServices and Rails: The Shit They Don't Tell You
43mRails services in the walled garden
33mServices and Rails: The Sh*t They Don't Tell You
23mAbstracting a Service - A Case Study in Forward Thinking with SOA