Temporal has released task queue priority and fairness features now generally available in both open source and Temporal Cloud. Priority segregates tasks into virtual queues (levels 1-5) so higher-priority tenants like paid users always execute before lower-priority ones. Fairness uses a weighted round-robin dispatch via fairness keys to prevent any single tenant from crowding out others on shared infrastructure. A live demo using a Python Streamlit app shows both features in action. Fairness is free in open source but a paid add-on in Temporal Cloud (0.1 action charge per action). The team also discussed future roadmap items including per-tenant dispatch rate limiting, concurrency controls, and token-based rate limiting for LLM workloads.
Sort: