Beyond Cloud Computing: Towards the Edge
summary
Claims that latency drops from ~120 ms to 1–5 ms when moving from typical cloud regions to nearby edge points, and that human thresholds sit around 100 ms, anchor the argument. Concrete tools named include Cloudflare Workers, Deno Deploy, AWS Lambda, Fly.io, Turso (replicated SQLite), Cockroach, Neon, PlanetScale, Redis, Workers KV. Notes: edge databases colocate query engines, offer transactional snapshots with weaker global consistency, and the fan‑out to distant cloud databases can multiply latency (example: Frankfurt–Sydney ~280 ms per round trip).
AI-written summary · about
chapters (26)
- 0:00Edge computing
- 2:25Outline
- 3:56Why latency matters?
- 5:43Network latency components
- 6:19Content delivery networks
- 8:50CDN challenges
- 9:11Cloud computing
- 11:20Cloud latency (cont'd)
- 11:52Edge latency
- 13:20Edge locations
- 13:37Near and far edge
- 14:38Edge platforms
- 15:37Edge + serverless
- 17:15Example: Cloudflare Workers
- 17:47Deploy once, run everywhere
- 18:06Example: AWS Lambda
- 19:10Key-value stores
- 20:52Cloud databases
- 22:29Latency compounds
- 23:07Smarter placement
- 24:02Edge databases TURSO
- 26:00No fanout problem with edge databases
- 26:32Data protection
- 27:25Edge-cloud interoperability
- 28:21Serverless API fragmentation
- 28:49Serverless costs
related talks
42mThe Edge at your fingertips, literally!
01h 24mThe Future of the Web: From Cloud to Edge
01h 06mServerless and Edge Computing in Practice
08mDefining the 'edge' in edge computing
46mFuture At The Edge Cloud Edge Devices: Large Scale Data Processing, Ops, And Challenges
18mScaling APIs to Billions: Extreme-Scale Reliability
29mJavaScript's Journey to the Edge
31mManaging Edge Computing with Serverless