Resource limitations and priority¶
Resource limitations¶
To protect our resources, all APIs have resource limits, which enqueue requests to avoid overloading the service.
If requests are not picked up for handling within a configured time limit (2 seconds), the request is rejected
with a 503 Service Unavailable.
All API clients are expected to handle the 503 response:
1) Clients should consider the request not handled, and retry the request 2) Clients should impose a delay before retrying, to avoid aggravating the load on the service.
Tip
A reasonable retry interval is 1 second, with exponential backoff.
API priority¶
To ensure timely response for interactive users and avoid unwanted impact of large bulk requests on other traffic,
we allow setting different priority on requests using a general header Argus-Priority.
By default, all requests are set to Argus-Priority: standard.
To separate bulk requests to a separate resource pool, set priority to bulk:
1 | |
API integrations which do heavy data loading are expected to set priority bulk
to avoid interrupting higher priority traffic.
Note
If heavy usage is observed, mnemonic may request integrating users to set priority bulk on their requests. For excessive use exceeding fair usage of our APIs, mnemonic may instruct users to change or disable their integration.
See Terms of Service