
AlexandreG.58943 (Customer) asked a question.
Hi,
We would like to use the Okta's AuthZ server to handle the API Access Control.
But we would like to know if there are some good practices.
Should we create 1 AuthZ server for our API Gateway and handle the Client ID's Scopes assignment using Access Policies? (i.e. this means that we could end up with a LOT of access policies for 1 AuthZ server - Is it an issue? Do you have some advises on the limits of the system)
Should we create mulitple AuthZ Servers to distribute the load generated by our API Gateway? Delivering 1 AutZ Server per API segment which will make the design more complex as we will have to duplicate some scopes and API subscriptions will have to be translated into multiple AuthZ Server requests.
Thank you in advance for your answer.

Hi Alexandre,
The recommendations are provided as a good grounding for utilising the authorisation server but will not fit all use cases. As such you may wish to review the API Rate limits found here: https://developer.okta.com/docs/api/getting_started/rate-limits. This highlights the rate limits that are applied,
Authorize request to a custom Authorization Server:
"/oauth2/{authServerId}/v1/authorize"
This has a rate limit ranges from 100 - 600 calls per minute based on the edition of Okta you are using, as such if you have multiple apps using the same authoirsation server this will be fine as long as the number of auth calls does not exceed that limit.
If you are able to work out the expected calls being made from your applications and the peak times if any then the above should help in identifying the best approach for your environment.