Skip to main content
Dulan Wirajith Lokunarangodage - Engineering articles author/ blog
HomeArticlesSeriesTopicsTipsAbout
Read latest
HomeArticlesSeriesTopicsTipsAbout
Read latest articles →
Dulan Wirajith Lokunarangodage - Engineering articles author/ blog
ArticlesSeriesTopicsRSS GitHub LinkedIn
© 2026 Dulan Wirajith Lokunarangodage. All rights reserved.
www.dulanwirajith.com
Home/Series/Realtime Infrastructure at Scale

Realtime Infrastructure at Scale

A behind-the-scenes series on rebuilding a live distributed system in production. No outages, real architecture decisions, and the mistakes we almost shipped.

5 parts · ordered reading path · next and previous navigation included
Start seriesJump to final partAll series
Reading progress
Start at Part 1 and move forward with the next button on each article.
Every article page now shows its series position, progress, and related navigation.
  1. Part 11/5
    Performance
    The Day Our Auction Platform Went Dark: A Production Post-Mortem
    A traffic spike brought down our entire real-time platform without warning. Here's the story of what broke, why it broke, and the architectural decisions that followed.
    Apr 29, 2026 · 5 min read · Part 1 of 5
  2. Part 22/5
    Architecture
    Redesigning for Scale: The Architecture That Replaced a Single Point of Failure
    After the production outage, we had a clear mandate: no more single points of failure. Here's the new distributed architecture we designed, from ALB through ECS Fargate to Redis, and why we made each choice.
    May 6, 2026 · 8 min read · Part 2 of 5
  3. Part 33/5
    Performance
    Connecting a Fleet: How Redis Pub/Sub Makes Socket.IO Scale Across Containers
    The core challenge of scaling Socket.IO is that each server instance is an isolated island. Here's how we used @socket.io/redis-adapter and the Socket.IO v4 upgrade to bridge those islands into one unified real-time system.
    May 13, 2026 · 8 min read · Part 3 of 5
  4. Part 44/5
    Architecture
    What Happens When a Container Dies Mid-Auction: Resilience Engineering in Practice
    Building a reliable system means planning for failure, not just success. Here's how we engineered Multi-AZ failover, sub-30-second crash recovery, and observability that turned hour-long incidents into 10-minute fixes.
    May 20, 2026 · 13 min read · Part 4 of 5
  5. Part 55/5
    Performance
    The Decisions, Near-Disasters, and 35% Cost Savings from Rebuilding a Live Bidding Engine
    Why Redis and not Kafka? Why Fargate and not Kubernetes? And what happened in staging that nearly derailed the entire migration? The honest debrief, tradeoffs, the mistake that almost broke production, and the final results.
    May 27, 2026 · 8 min read · Part 5 of 5