Rate limiting

This limiter endpoint can be used to control the rate that incoming initial requests received from the network are accepted for processing by the SIS.

The SIS requests one unit of work from this limiter endpoint for each initial request that it receives. If the request is granted then initial request processing proceeds normally. If, however, the work request is rejected by a rate limiter connected to the endpoint, then the SIS stops processing the initial request and returns an error response determined by the rate-limited error response status code and Retry-After period configuration properties.

This limiter endpoint can be used to control the rate that incoming initial INVITE requests received from the network are accepted for processing by the SIS.

This limiter endpoint is a "sub-endpoint" of the inbound endpoint. For each initial INVITE, the SIS requests one unit of work from the inbound endpoint, and if that is granted, another unit of work is requested from the inbound_invites endpoint. By configuring the inbound endpoint with a rate limiter of higher capacity than inbound_invites, the SIS can throttle initial INVITEs while still allowing other initial requests to proceed.

If the initial INVITE is rejected by a rate limiter connected to the endpoint, then the SIS stops processing the INVITE and returns an error response determined by the rate-limited error response status code and Retry-After period configuration properties.

This limiter endpoint is used by the SIS when Rhino key/value store replication is in use and the SIS receives a mid-dialog request on a dialog that it has no local state for.

Before querying the Session Ownership Facility in an attempt to retrieve the state for the dialog, the SIS requests one unit of work from this limiter endpoint. If the request is granted then the Session Ownership Facility is queried normally. If, however, the work request is rejected by a rate limiter connected to the endpoint, then the SIS stops processing the mid-dialog request and returns an error response determined by the rate-limited error response status code and Retry-After period configuration properties.

In summary, the purpose of this limiter endpoint is to limit the number of queries made to the key/value store backend database (typically Cassandra) under failover conditions.