Update: Usage Plan values for the Transfers API v2024-06-01 initiatePayout operation
Amazon has lowered the Transfers API v2024-06-01 initiatePayout throttling to an average of 0.017 req/s (≈1 call per minute) with a burst limit of only two calls. The new limits trigger 429 errors and include a Retry‑After header, forcing sellers to add at least a 60‑second pause or queue‑based pacing.
Overview
Amazon has lowered the throttling limits for the initiatePayout call in the Transfers API version 2024‑06‑01. The new policy caps the average request rate at roughly one payout per minute and reduces the burst allowance to two calls. Sellers that run automated payout workflows need to re‑engineer their integrations now to avoid 429 “Too Many Requests” errors that could postpone customer disbursements.
Key Points
- Average rate reduced to 0.017 req/s — The API now allows about one payout request every 60 seconds, a sharp drop from the former 0.5 req/s (≈30 requests per minute).
- Burst window cut to two calls — Only two consecutive payout submissions can be made before the system starts enforcing the average‑rate rule, compared with a previous burst capacity of 30 calls.
- 429 errors become common — Applications that previously sent batches of payouts back‑to‑back will now receive “Too Many Requests” responses far more often, especially when the third request arrives within the same second.
- Potential delay of several minutes per batch — Sellers who relied on near‑real‑time payouts may experience latency that adds up to a few minutes for each group of transactions, affecting cash‑flow timing.
- Mandatory pacing or queuing logic — To stay within the new limits, developers must insert deliberate pauses between calls or route requests through a throttling‑aware queue.
- Retry‑After header guides back‑off — When throttling occurs, the response includes a
Retry-Aftervalue that tells the client exactly how long to wait before retrying, simplifying automated back‑off strategies.
How the Updated Usage Plan Works
- Rate‑window calculation — The service computes the average number of initiatePayout calls over a rolling time window. For instance, if a seller’s system fires 12 requests in ten minutes, the average equals 0.02 req/s, which exceeds the 0.017 req/s ceiling and triggers throttling.
- Burst enforcement — The platform permits only two immediate calls before applying the average‑rate check. A practical example: a seller sends two payout requests one after another; a third request issued within the same second is rejected with a 429 status even though the ten‑minute average might still be under the limit.
Analysis & Recommendations
Why This Matters
The tighter limits mean automated payout batches will now hit 429 errors after the second call, adding several minutes of latency per batch and potentially delaying customer refunds. Sellers must redesign their payout logic to avoid cash‑flow gaps and comply with the new average‑rate rule.
Key Takeaways
- Average rate cut to 0.017 req/s (≈1 payout per minute) from the previous 0.5 req/s.
- Burst capacity reduced from 30 calls to only 2 immediate calls.
- 429 responses now include a Retry‑After header that typically specifies a 60‑second wait.
- A batch of 30 payouts would now take ~30 minutes instead of a few seconds under the old limits.
Recommended Actions
- →Modify your payout scheduler code (e.g., in the service that calls initiatePayout) to insert a minimum 60‑second delay between calls.
- →Set up an Amazon SQS queue (AWS Console > SQS > Create Queue) and have a worker poll it, releasing at most two calls per second and one per minute ...
- →Implement retry logic that parses the Retry‑After header from 429 responses and retries after the indicated delay, applying exponential back‑off fo...
Comments
Join the discussion
Log in or create an account to share your thoughts on this update.
No comments yet. Be the first to share your thoughts!