| This thesis examines and suggests solutions to some of the technical problems and performance problems encountered when designing modeling, and evaluating a Service Control Point (SCP) architecture. The SCP is known as the "heart" of the Intelligent Network (IN). It is essentially a transaction processor-based computer system designed to execute IN services under stringent performance requirements and Quality of Service (QoS). Expected future problems for the SCP are mainly due to the introduction and increasing complexity of new, and already deployed IN services (e.g., freephone service, calling card service, televoting, virtual private network, universal access number, etc.). The challenges in designing the SCP architecture will involve issues such as performance, evaluation of the system, quality of service definition, estimating the capacity and delays/blocking, traffic rate measurements, availability, reliability, flow control, management activities, interworking, etc.;We present and study an SCP architecture by means of queuing models and simulations. Our study differs from previous research as we consider in detail the internal SCP architecture. The proposed SCP architecture is designed as a multiprocessor-based system. It provides scalability while maintaining the requirements of the three most important performance metrics: SCP throughput, SCP response time and SCP call sampling rate. Scalability insures that IN services and SCP call sampling growth rates can be handled without impacting the performance of the SCP. The architecture also lends itself to discuss a multitude of performance and modeling issues such as service logic program storage, IN call processing activity, real-time traffic rate, main memory constraints, blocking probability, buffer dimensioning, bottleneck problems, management and administrative activities, statistical distributions, and specialized processors. |
