eroman examines how Azure Functions for Python apps benefit from native orjson support, showing measurable performance improvements in latency and throughput for serverless solutions.

Scaling Azure Functions Python with orjson

Introduction

Delivering high-performance serverless applications is essential for Azure Functions, which powers workloads like APIs, event processing, and automation. A major contributor to overall performance is JSON serialization and deserialization for Python-based functions, whether they’re interfacing via HTTP, Event Hubs, Service Bus, or Cosmos DB. Recognizing this, Microsoft has introduced support for orjson—a fast, Rust-based JSON library—in the Python worker for Azure Functions.

What is orjson, and Why Does It Matter?

orjson is a performance-oriented library for JSON serialization/deserialization, leveraging Rust for memory efficiency and speed.
Azure Functions Python Library now auto-detects orjson: if it’s found in your requirements.txt or virtual environment, Azure Functions switches to orjson, bypassing the slower standard Python json module. This improvement requires no code change in your apps.
JSON serialization is critical for any serverless function processing messages, API payloads, or integrating event-driven Azure services.

Why orjson Beats Traditional Libraries

Written in Rust for optimized memory handling
2-4x faster than ujson, up to 10x faster than json for large or deeply nested payloads
Generates more compact output—useful for bandwidth-sensitive applications
Strict RFC 8259 compliance for interoperability

Measured Performance Gains

HTTP and Event Hub Benchmarks

Locally:

HTTP: 10,000 requests (20KB payload)
- json: 10:01 min total, avg request 3.08s, median 2.16s, dropped: 289
- orjson: 6:01 min total, avg request 1.8s, median 1.51s, dropped: 0
Event Hub: 1,000 messages (20KB)
- json: 4:11 min, 3.98 messages/sec
- orjson: 2:50 min, 5.88 messages/sec

Cloud Results (Azure Functions Flex Consumption):

100 instances, 2GB RAM, Azure Load Testing for throughput (req/s) and latency (ms)
At 1KB to 992KB JSON payloads, orjson improved throughput by up to 6% and reduced average response time by up to 400ms
For Service Bus, orjson showed +38% message rate at 1,000 messages, up to +15% at 10,000 messages

Integration Challenges & Lessons

Compatibility across supported Python versions (3.9–3.13) required robust testing
Conditional loading means orjson is only used if present; otherwise, fallback to json
Accurate benchmarks needed to account for cold starts, payload variance, and cloud transience

Success Stories

Up to 40% reduction in average request times with larger payloads
Event Hub scenarios: up to 50% higher message throughput
Drop-in improvement by merely adding orjson as a dependency, no code refactoring required

Next Steps for Developers

Review Azure Functions Python Guide for best practices
Consider adding orjson to requirements.txt for immediate performance upgrades
Monitor upcoming enhancements for deeper worker-level optimizations

Conclusion

Native orjson support in Azure Functions for Python apps offers developers fast, easy performance benefits for JSON-heavy serverless workloads. Results are both significant and practical for production scenarios, especially for applications with high message volume or large payloads.