Live Polling at Viral Scale
System Design diagram · made in Skemio

What it shows
- Clients buffer votes locally for 2-5s to eliminate hot HTTP request floods.
- Redis counter shards split vote tallies across keys to avoid hot-key stalls.
- Async aggregation workers flush batched snapshots to PostgreSQL every 5s.
The takeaway
Absorb spikes by batching at the client and partitioning hot counters in memory.
- #SystemDesign
- #Redis
- #Backend
- #Scalability
- #SoftwareEngineering





