Take an expert-level deep dive into the core architecture and execution model that powers Azure Durable Functions. In this session, Florian Lenz will explore how the Durable Task Framework uses event sourcing to persist orchestration state, replay history, and checkpoint activities to ensure reliable, long-running workflows. You’ll gain insight into the deterministic replay engine, learn why activities must be idempotent, and see how race conditions are avoided during state reconstruction.

We’ll pull back the curtain on the backend—examining the Durable Task Scheduler’s role in queuing and distributing work items, how it interacts for event buffering, and the mechanics of orchestrator and activity worker coordination. You’ll understand how the Functions scale controller leverages the scheduler and storage metrics to scale out orchestrations safely, and how to tune checkpoint frequency, throttle fan‐out, and manage storage throughput to optimize performance and cost.

Finally, we will cover real-world troubleshooting strategies for stuck orchestrations, runaway replay loops, and excessive storage costs. By the end of this session, you’ll have a clear mental model of Durable Functions under the covers and be equipped to build, monitor, and optimize resilient, high-scale workflows in Azure.