Efficiently providing multiple grades of service with protection against overloads in shared resources

A method of controlling admission of new customers to a shared resource, in which said new customers can be provided with multiple grades of service, including protection against overloads from existing customers and other new customers, said method comprising the steps of

• retrieving stored information defining the requirements of said existing customers, and obtaining the requirements and grade of service desired by each of said new customers;
• responsive to said retrieved information, determining if said requirements of said new customer can be satisfied without violating the requirements of said existing customers, said determination being made in accordance with a product-form resource sharing model which has a normalization constant;
• and
• allowing said new customer to be admitted to said shared resource with said described grade of service only if said requirements of said new customer can be satisfied,
• wherein said determining step includes calculating blocking probabilities for usage of said shared resource by said new customers and said existing customers by numerically inverting a generating function of said normalization constant of said resource sharing mode.

A method of controlling admission of a new customer to a shared resource serving existing customers, in which said existing customers and said new customer can be provided with multiple grades of service, each of said grades of service being defined by a traffic requirement and a blocking requirement, said method comprising the steps of

• obtaining information indicating (a) the grade of service of said existing customers and (b) the grade of service desired by said new customer;
• responsive to said grade of service desired by said new customer, assigning upper-limit UL and guaranteed-minimum GM bounds of the usage of said shared resource to said new customer, said upper-limit bound putting an upper limit on the usage of said shared resource by the customer and said guaranteed-minimum bound representing a guaranteed minimum usage of the shared resource,
• determining if said grade of service of said new customer can be provided without violating the requirements of said existing customers, and
• if said grade of service of said new customer can be provided, allowing said new customer to be admitted to said shared resource with said desired grade of service,
• wherein said determining step includes computing a blocking probability for said new customer by numerically inverting a generating function of a normalization constant of a resource sharing model, and determining if said blocking probability is less than the blocking requirement and satisfies the UL and GM bounds assigned to said new customer.

A method for real time admission control to determine in real time whether a prospective new customer can be admitted to a resource with a desired grade of service, comprising the steps of

• determining the desired grade of service for the prospective customer,
• determining if the desired grade of service can be met considering both the prospective customer and all existing customers, wherein said determining includes computing blocking probabilities for said prospective customer and said existing customers by numerically inverting a generating function of a normalization constant of a resource sharing model,
• if the desired grade of service can be provided, admitting the new customer with the desired grade of service, and
• if the desired grade of service cannot be met, then determining if a lower grade of service is feasible.

A method of adjusting the capacity of a shared resource so as to satisfy new levels of customer demand, comprising the steps of

• creating a resource sharing model for the shared resource that has a product-form steady-state distribution and a normalization constant,
• responsive to a change of customer demand, determining the new demand level for said shared resource, and
• adjusting the capacity of the shared resource to satisfy blocking requirements for all customers,
• wherein said adjusting step includes calculating blocking probabilities for usage of shared resource by said customers by numerically inverting a generating function of said normalization constant of said resource sharing model.

A method of diverting customer usage from failed resources to alternative resources in a shared resource system, comprising the steps of

• monitoring the system to determine failure of a resource,
• identifying appropriate alternative resources, and
• adjusting usage to meet blocking probability requirements of existing and diverted customers on said alternative resources,
• wherein said adjusting step is made in accordance with a product-form resource sharing model which has a normalization constant, and
• wherein the blocking probabilities for usage of said shared resource by said diverted customers and said existing customers are calculated by numerically inverting a generating function of said normalization constant of said resource sharing model.

A technique for providing different grades of service and protection against overloads to customers sharing a resource, comprising the steps of

• assigning each customer upper-limit (UL) and guaranteed-minimum (GM) bounds on requests, said upper limit bound putting an upper limit on the number of requests from that customer that can be in service at any time, and said guaranteed-minimum bound guaranteeing that there will always be available resource units in the resources to serve a specified number of requests from that customer, and creating a resource-sharing model having a product-form steady-state distribution, and
• solving the resource-sharing model using a blocking probability computer (BPC) which operates by
• (a) directly expressing said model in terms of normalization constants appearing in the product-form steady-state distribution,
• (b) computing the normalization constants by constructing a generating function of the normalizing constant and
• (c) numerically inverting the generating function.

A method of controlling admission of new customers to a shared resource that is serving existing customers, said method comprising the steps of

• determining the blocking probability requirements of said existing customers;
• determining the blocking probability requirements and grade of service desired by each of said new customers;
• determining if said requirements of the new customers can be satisfied without violating said requirements of said existing customers, said determination being made in accordance with a product-form resource sharing model which has a normalization constant, and said determination includes calculating the blocking probabilities for usage of said shared resource by said new customers and said existing customers by numerically inverting a generating function of said normalization constant of said resource sharing model;
• and
• if said new customers requirements can be satisfied, allowing said new customer to be admitted to said shared resource.