API Limits Documentation / Throttling

Question

Will the API limits, especially around things like per-API-key throttling, be documented?

The specific use case I'm thinking about is around managing App Store Connect access for users inside of large organizations. Suppose there exists a system that is responsible for facilitating user invitations, managing user roles, and periodically auditing that user access is valid according to internal policies.

Since such a system would have both periodic API usage (for auditing) and human-generated API usage (for invitations and roles), would API requests made with the same API Key be at risk for getting throttled?

Would it be wise to have a separate API key for requests that we can control the rate for?

For human-generated requests, would I need to implement some sort of least-recently-used API Key pool to avoid getting throttled?

These sorts of questions (and more) could be self-solved if the API documentation were to describe the how the API does throttling.

App Store Connect API

6.6k

Posted by

Calypso

Reply

Add a Comment

Answer 1

Hi Calypso!

I can't address your question about specific limits, which may vary. But I can provide some technical details that may help in your planning.

Every response from the API includes a

X-Rate-Limit

HTTP header. Its value has this form:

user-hour-lim:10;user-hour-rem:8;

user-hour-lim

indicates the number of requests you can make per hour with the key being used.

user-hour-rem

shows how many requests remain. So in this example, you are limited to 10 requests per hour, and you have 8 remaing. (Note that actual limits will be higher than this example.) This is a "rolling hour" so in practice, at any moment, the

user-hour-rem

value is your per-hour limit minus the total requests you've made in the previous 60 minutes. The

user-hour-lim

may change over time, so if you're trying to throttle a long-running process to work continuously you should consult this header on your first request and set your delays accordingly.

If you exceed your per-hour limit, the API will begin rejecting requests with an HTTP 429 response with the error code

RATE_LIMIT_EXCEEDED

. You can find out more about how to interpret API errors here:

https://developer.apple.com/documentation/appstoreconnectapi/interpreting_and_handling_errors

If the limit in place for your keys is too small for your user-initiated scenarios, you might consider caching response data, coalescing requests from multiple users into one, or implementing a queue to fulfill user requests with throttling.

Hope this helps!

Posted by

App Store Connect Engineer

Add a Comment

Answer 2

Hi Calypso!

I can't address your question about specific limits, which may vary. But I can provide some technical details that may help in your planning.

Every response from the API includes a

X-Rate-Limit

HTTP header. Its value has this form:

user-hour-lim:10;user-hour-rem:8;

user-hour-lim

indicates the number of requests you can make per hour with the key being used.

user-hour-rem

shows how many requests remain. So in this example, you are limited to 10 requests per hour, and you have 8 remaing. (Note that actual limits will be higher than this example.) This is a "rolling hour" so in practice, at any moment, the

user-hour-rem

value is your per-hour limit minus the total requests you've made in the previous 60 minutes. The

user-hour-lim

may change over time, so if you're trying to throttle a long-running process to work continuously you should consult this header on your first request and set your delays accordingly.

If you exceed your per-hour limit, the API will begin rejecting requests with an HTTP 429 response with the error code

RATE_LIMIT_EXCEEDED

. You can find out more about how to interpret API errors here:

https://developer.apple.com/documentation/appstoreconnectapi/interpreting_and_handling_errors

If the limit in place for your keys is too small for your user-initiated scenarios, you might consider caching response data, coalescing requests from multiple users into one, or implementing a queue to fulfill user requests with throttling.

Hope this helps!

Posted by

App Store Connect Engineer

Add a Comment

API Limits Documentation / Throttling

Accepted Reply

Replies