TABLE OF CONTENTS The foregoing discussion of the nature of overloads and their impact on network resources serves to motivate the requirements discussed in this section.
A clear specification of how an overload control should behave (i.e. a set of overload control requirements) is essential before deciding how it may best be implemented to achieve that behaviour (the latter aspect is discussed in detail in section 7.3.6).
The derivation of a set of overload control requirements is discussed here in the context of a resource (called the target) which is overloaded by service requests coming solely from a set of resources where restriction is applied (known as points of restriction , or restriction points ). (This assumes that all overload sources are potential restriction points.) A point of restriction may be located at a neighbouring sending node (as is the case with ISUP ACC), or at the originating node for a service request (as is the case with the private network-network interface (PNNI) protocol when the originating node and the target lie in the same peer group), or possibly somewhere in between the two. The target is assumed to have an internal mechanism to detect if it is in overload and an admission control which decides whether each individual service request is accepted by the target or not. It is further assumed that if the admission control at the target resource decides not to accept a service request, then it deals with it by one of two mechanisms:
either by...
