The number of connections and requests are limited to ensure that a single user (or bot) cannot overload the whole server.
50 connections / address
50 requests / second / address
150 requests / second (burst)
>150 requests / second / address (access limited)
With this configuration, a particular visitor can open up to 50 concurrent connections and issue up to 50 requests / second.
If the visitor issues more than 50 request / second, those requests are delayed and other clients are served first.
If the visitor issues more than 150 request / second, those requests will not processed anymore, but answered with the 503 status code.
To adjust this limits for special applications such as API calls, set a higher load zone in your Website Level nginx configuration:
# connection limits (e.g. 75 connections) limit_conn addr 75; # limit requests / second: (small, medium, large) limit_req zone=medium burst=500; limit_req zone=large burst=1500;
To apply the changes reload the nginx configuration with the
small = 50 requests / second (burst: 150req/sec)
medium = 150 requests / second (burst: 500 req/sec)
large = 500 requests / second (burst: 1500 req/sec)
Note: the default zone is “small” and will fit most use cases
For details, see the nginx Module ngx_http_limit_req_module documentation.