Thairath Livescore (Realtime API)

📚 Overview

• Handle

  high concurrency during live matches

• Deliver

  low-latency, bandwidth-efficient updates

• Provide

  fine-grained subscription-based data streaming

🚀 Core Architecture

🔄 Data Pipeline

• External partner APIs →

  data scraping every 15 seconds

  (rate-limited)

• Data processed and published into

  Redis Pub/Sub

• Messages encoded using

  MessagePack (compressed binary format)

• Distributed to gateway pods → decompressed → streamed to clients via SSE

⚙️ Infrastructure & Scaling

• Kubernetes-based deployment (horizontal scaling)

• Each pod:

• ~

  0.5 CPU (500m)

  +

  512MB RAM

• Handles ~

  6,000+ concurrent SSE connections per pod

1
500m CPU = 0.5 CPU core

• Auto-scaling strategy

  :

• Scale pods horizontally based on traffic

• Maintain stable performance under heavy load

🚀 Key Features

📡 Real-Time Streaming (SSE)

• Lightweight, persistent connection for pushing live updates

• Optimized for high concurrency

📦 MessagePack Compression

• Reduces payload size significantly vs JSON

• Improves throughput and latency

🔁 Redis Pub/Sub Distribution

• Decouples ingestion from delivery

• Enables multi-node horizontal scaling

⚙️ Advanced System Capabilities

🧠 Auto Compare (Delta Updates)

• Clients receive

  only changed data

• Reduces unnecessary bandwidth usage

🧾 Version-Based Sync

• Clients send current version → receive only diffs

• Ensures efficient state synchronization

📡 Channel-Based Subscription

• Fine-grained topics like:

  1
  sport-livescore/{matchId}

🌐 Wildcard Channels

• Supports patterns:

  1
  sport-livescore/*

❤️ Auto Heartbeat

• Sent every

  15–30 seconds

• Keeps connection alive + detects disconnects

🚀 Initial Data Sync

• On first connect:

• Sends latest full state

• Aligns user with current system version

🧩 Channel Mapping System

• Dynamic channel creation

• Flexible routing between publishers and subscribers

🛠️ Tech Stack

• Backend:

  Node.js (Realtime API Gateway)

• Realtime Layer:

  Redis (Pub/Sub)

• Protocol:

  SSE (Server-Sent Events)

• Data Format:

  MessagePack

• Infrastructure:

  Kubernetes (auto-scaling pods)

👨‍💻 Responsibilities

• Designed and built the

  entire real-time gateway architecture

• Implemented

  data ingestion pipeline (15s scraping from partner APIs)

• Built

  Redis-based pub/sub messaging system

• Designed

  MessagePack encoding/decoding pipeline

• Developed

  SSE streaming layer for real-time delivery

• Implemented advanced features:

• Delta updates (auto compare)

• Version-based syncing

• Channel subscription system (with wildcard support)

• Heartbeat & connection management

• Deployed and optimized system on

  Kubernetes with horizontal scaling

🔮 Future Scalability

• Supports

  multi-source ingestion (multiple providers)

• Can serve

  multiple client types

  (web, mobile, third-party APIs)

• Designed as a

  reusable real-time infrastructure platform

🎯 Impact

• Delivered

  low-latency real-time experience

  at scale

• Reduced bandwidth via

  delta updates + MessagePack compression

• Achieved

  high concurrency (~6,000+ users per pod)

• Built a

  reusable real-time system for future products