The invention discloses a content replica placement method and a content replica placement system used in a content delivery network (CDN). According to the content replica placement method, a chosen replica placement node needs to meet the service response time requirement primarily. For any node Mi in the CDN, when the node Mi serves as an access node, preparatory replica placement node sets Ci={Ni1, Ni2, -Nik} which enable the node Mi to meet the service response time requirement and the local-storage requirement are found. A preparatory node Ns which has highest frequency of occurrence in the preparatory replica placement node sets is found from all the preparatory replica placement node sets. For each preparatory replica placement node set Ci which comprises the preparatory node Ns, the total number of requests of a corresponding access node Mi is calculated, and if the number is smaller than the maximum load request number of the preparatory node Ns, the node is chosen as an actual replica placement node, and otherwise a node which has second highest frequency of occurrence is considered. According to the content replica placement method, upper limit of load of each node of the CDN is limited based on that the node meets the servicing response time requirement, and simultaneously content replica placement quantity is reduced, and expense is optimized and updated.