API Usage, Rate Limits & Quotas: Token Buckets, Retries and Cost

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Convert a client's per-minute request log into requests per second, find the peak minute, count the minutes over the 600-per-minute limit and the excess requests.
• Simulate a token bucket with capacity 10 and a refill of 4 tokens per second for 12 seconds of traffic and measure how many requests it rejects.
• Run the same 12 requests through a fixed-window and a sliding-window limiter (5 per 60 s) and count what each one rejects.
• Build a six-retry schedule: exponential delays, a cap, full jitter from given random draws and the total time a client waits before giving up.
• Measure Routely's daily 429 rate, how many requests still failed after retries, the week's final success rate, and model what retries add in success and in load.
• Calculate the overage of four plan tiers, Quillstack's monthly overage revenue, the total monthly call volume and the peak requests per second it implies.
• Work out how many API calls Routely saves with a cache and with batch requests, and what that saves on the monthly bill.
• Calculate the cost to serve an average customer of each plan, the gross margin per tier, the break-even usage of the Starter plan, and chart the margins.