Ember at Scale
summary
Ember scales to hundreds of engineers and ~400 million users by shipping a server-side component called the base page renderer that runs the same Ember app in Node, streams chunked HTML with a BigPipe pattern, proxies API responses into index.html via ember prefetch, and uses a sidecar deploy so JVM services keep existing networking. Production use for months yields 1–2s faster initial renders, and crumb‑trace v8 profiling exposes actual CPU and eval hotspots for followup optimization.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:45EMBER AT SCALE
- 1:04THIS IS OUR STORY
- 1:252013
- 1:57PRODUCT & TECH BALANCE
- 3:13If it ain't broken don't fix it.
- 3:492014
- 4:16SOFTWARE ROT
- 4:59UNIFY & REDUCE COMPLEXITY
- 6:07CONSIDERATIONS
- 6:20CENTRALIZED SUPPORT
- 7:45LESS IS MORE
- 8:43CHANGE IS THE CONSTANT
- 10:16SOLVING PROBLEMS
- 10:48INITIAL RENDER
- 11:15BASE PAGE RENDERER (BPR)
- 12:47BIG PIPE
- 16:58npm install ember-prefetch
- 19:48ember fastboot --resource-discovery
- 20:16SERVING THE APP
- 24:04DEPLOYMENTS
- 25:54npm install chrome-tracing
related talks
36mPanel: Ember at Scale
01h 01mOpening Keynote
22mUsing Ember to Make the Seemingly Impossible Easy
01h 10mPanel: the future of Ember
35mPanel: Ember – Why and Where?
22mUsing Ember to Make the Seemingly Impossible Easy
48mInside Glimmer: What Makes Ember's Rendering Engine Tick
46mArt and Science of Shipping Single Page Apps with Ember.js