Dealing with Unreliable Networks

on YouTube ↗29m

" summary

Mobile networks are slow and flaky worldwide, so the app adds concrete fixes: retry with exponential backoff reduces transient failures — measured up to a 80% drop in some markets — combined with idempotent GUIDs or server-side dedupe to avoid duplicate posts. Important actions are written to disk before send for resiliency. Performance gains come from POPs/CDN, moving SSL termination, adopting HTTP/2 prioritization and using WebP on Android (30–50% smaller) or progressive/low‑res placeholders to cut initial image render time by ~75%. A network simulator and client-side metrics prove and tune these changes.

AI-written summary · about

" chapters (19)

1. 0:00Intro
2. 0:34Dealing with Unreliable Networks
3. 1:34Reliability
4. 2:46What if that network call fails?
5. 4:10Android
6. 6:43Resiliency
7. 7:59Performance
8. 8:56Points of Presence (POPs)
9. 10:04HTTP/2 & SPDY
10. 11:30WebP
11. 13:34Progressive JPEG
12. 14:54Low-res Images
13. 15:46Waiting for requests
14. 17:14Request Prioritization
15. 21:38Network Forecasting
16. 22:57Empathy
17. 24:43Grey Box Times
18. 25:52Network Debugging
19. 27:04Network Simulation

" related talks

• 52mPanel 2: NetworkAndy Davies, Ilya Grigorik, John Cleveley, Steve Thair, Jackson Gabbard · Edge Conference 2013
• 01h 14mOptimizing App Performance on Slow NetworksAsya Sviridenko · Mobius 2020 Piter
• 39mMobile Networking Challenges - Networking @ScalePeter Griess · Networking @Scale 2015
• 14m89 characters of base-11?! Mobile networking in rural Ethiopia!Ben Kuhn · !!Con 2020
• 44mHTTP in a Hostile WorldJesse Wilson · droidcon Toronto 2019
• 11mIs the Internet Reliable Yet?Robert Peters · Velocity Conference New York 2015
• 35mIs your app really connected?Piotr Wittchen · Mobilization 8
• 16m"Latency: Why You Should Worry and What You Can Do About It"Philip Tellis · Velocity 2010