The publish/subscribe (pub/sub) system has a wide potential application due to its asynchronous, many-to-many and loosely-coupled communication properties. However the existing pub/sub systems can't satisfy the delay requirement of application in dynamic distributed computing environment. In order to solve this problem, this paper extends the syntax of pub/sub system, sets up the delay model, proposes an earning-based time-constraint assurance technique for distributed pub/sub system, and puts forward a scheduling algorithm named as MTEP (maximum total earning priority). The algorithm satisfies the specified delay requirement for both publisher and subscriber, makes use of available bandwidth efficiently with the price and penalty information agreed with subscribers, and adapts to the dynamic network environment. Experimental results show that the strategy can make subscribers receive many more valid events and improve system earning significantly comparing with the classical FCFS, fixed priority and least remain time priority algorithms.