Rate Limits​, Timeouts and Retries

Rate limits

The rate limit is tied to your IP address and allows 1,000 requests per minute. This limit applies separately to each environment—sandbox and live. If you exceed this limit, you'll receive responses with an HTTP 429 status code and a relevant error message.

Note: If you're using Midaz open source, you have the flexibility to adjust the rate limit parameters directly from the Helm file. For detailed instructions on how to make these adjustments, please refer to the Customizing your open source instance section in the documentation page.

Retries

API requests may fail due to various reasons such as network issues, API rate limits, timeouts, or service incidents. As a best practice, we recommend incorporating a request retry mechanism with the following attributes:

1. Only retry requests that failed with the following HTTP status codes:

• 5xx (Server Errors)

• 429 (Rate Limits)

• 408 (Timeouts)

2. Implement retries with exponential backoff and/or jitter.

3. Ensure to use an idempotency key where applicable.

Timeouts

All of our APIs are designed with timeouts to ensure rapid failure and allow for request retries. It's a good practice to set similar timeouts on the client side as well.

For the SaaS model, our APIs enforce a timeout of 29 seconds.

Note: If you're using Midaz open source, you have the flexibility to adjust the timeout parameters directly from the Helm file. For detailed instructions on how to make these adjustments, please refer to the Customizing your open source instance section in the documentation page.