Powering Adobe's Real-Time Advertising Analytics with gRPC Streaming
summary
Adobe’s Performance Insights backend replaces heavy REST/JSON responses with server-side streaming over gRPC, using Protobuf binary encoding and per-endpoint concurrency and deadlines. The reporting service streams column headers and rows incrementally, cuts request size (39% smaller payloads vs JSON, up to 72% bandwidth reduction), reduces TCP connections fivefold, and drops server memory by ~5x; streaming yields ~37% faster responses (3000ms → 1880ms) and real production wins.
AI-written summary · about
chapters (10)
- 0:00Introduction to real-time ads
- 2:30Campaign structure overview
- 7:20Backend architecture with gRPC
- 9:08Why switch from REST to gRPC
- 13:44Binary encoding benefits
- 16:56Legacy REST vs gRPC comparison
- 19:23Performance metrics analysis
- 21:51Bandwidth and latency gains
- 22:55Architecture impact summary
- 23:15Conclusion and Q&A
related talks
23mAdobe + Aerospike: Real-time Streaming Segmentation
28mFrom Backend Powerhouses to Web-Friendly APIs: Unlocking
41mLessons Learned Implementing a Resilient, Distributed, Real-Time Architecture
01h 03mMonitoring Performance on Massive Games.
44mAdvertising at the Speed of Light (DMP Platform)
26mAdopting gRPC for Low Latency, High Bandwidth Scenarios
36mHow Adobe is Optimizing Resource Usage in Kubernetes
41m8K Codec Low-Latency Livestreaming, and AI Streaming