SYSTEM AND METHOD FOR INTELLIGENT LOAD DISTRIBUTION TO MINIMIZE RESPONSE TIME FOR WEB CONTENT ACCESS

A content delivery system having m servers, S′={S1, . . . ,Sm}, n active customers, C′={C1, . . . , Cn}, and g geographic locations, G′={G1, . . . , Gg} is disclosed, wherein sdelk is a server delay of server Sk, ndelj,k is a network delay observed by customers in geographic location Gj while retrieving content from server Sk, pj is a priority value for customer Ci, ci is a total load of customer Cii, ui,j is a fraction of requests coming to customer Ci from region Gj, ai,j,k is a mapping representing a fraction of requests coming to customer Ci from region Gj that have been redirected to server Sk, and sk represents a load capacity of server Sk. Within such a system, a method for distributing server loads includes the steps of representing an average prioritized observed response time asand then generating a mapping that assigns requests from customers to a particular server while minimizing AORT. A heuristic algorithm is used to generate the mapping, wherein large ai,j,k values are assigned to small ui,j×ci×(sdelk+ndelj,k) values to produce a smaller overall AORT value.