Rate-limit downstream APIs with separate Task Queues
Avoid HTTP 429 errors from APIs such as Stripe, OpenAI, and Twilio by giving each API its own Task Queue with a queue-wide Activity rate limit, in Python.
3 docs tagged with "Routing"
View all tagsRoute GPU and high-memory Activities to dedicated Workers
Give GPU, high-memory, and CPU Activities their own Task Queues so each runs only on Workers with the right hardware, shown with a Python ML pipeline.
Run all of a Workflow's Activities on the same Worker
Keep downloaded files, loaded ML models, and open connections on one machine by routing each Activity to a Worker-specific Task Queue, with Python code.