How to build a social network entirely on serverless

on YouTube ↗33m

" summary

Yubl migrates from a monolith to serverless by publishing state changes to Kinesis and wiring Kinesis + Lambda + API Gateway for search, feeds, recommendations, push and BI; migration finishes in about six months for a million users. Resulting system runs around 200 functions, jumps to 80–100 deployments/month, uses BigQuery/Athena, SNS, Redis and Neo4j for graphs, and replaces a $3,000/month Mixpanel bill with cent‑level per‑query costs.

AI-written summary · about

" chapters (14)

1. 0:00Introduction
2. 0:42Hubble
3. 1:41Why
4. 8:13Events
5. 8:46Microservices
6. 9:35cascade failures
7. 10:28eventdriven architecture
8. 11:43workflow engine
9. 14:08one repo per function
10. 14:47one repo per microservice
11. 17:56migration pattern
12. 20:22cost saving
13. 27:13recommendation feature
14. 30:13more scalable

" related talks

• 29mHow to build a social network on serverlessYan Cui · apidays Barcelona 2019
• 44mHow to build a social network entirely on serverlessYan Cui · NDC Porto 2019
• 24mServerless gives you wingsYan Cui · Frontend Love 2020
• 39mServerless in production, an experience reportYan Cui · Codemotion Milan 2017
• 33mBuild a social network with Serverless and GraphQLYan Cui · JSWorld Conference 2021
• 01h 02mServerless in production, an experience reportYan Cui, Domas Lasauskas · NDC London 2018
• 01h 02mFrom Monolith to ServerlessRajpal Singh Wilkhu · NDC Oslo 2017
• 30mThinking Serverless: From User Request to Serverless SolutionJames Beswick · GOTO Amsterdam 2022