In this paper, we design and implement network-based Latency EQualization (LEQ), which is a service that Internet service providers (ISPs) can provide for various interactive network applications. Our LEQ architecture provides a flexible routing framework that enables the network provider to implement different delay and delay difference optimization policies in order to meet the requirements of different types of interactive applications. To achieve LEQ routing, we formulate the hub selection problem, which decides which routers in the network can be used as hubs and the assignment of hubs to different client edge routers to minimize delay difference. The LEQ architecture contains 3 components LEQ Service Manager, Hubs and Router.