In the mobile edge computing (MEC) environment, tasks can be offloaded to edge servers for better service. However, the locations of users are not fixed. As users move, tasks need to be migrated to the vicinity of users to ensure quality of service (QoS). At present, most research considers the problem of service migration from the perspective of a single user. The single user migration scheme is only concerned with it's own optimal choice, without considering the interference between multiple users, which will lead to multiple users blindly competing for resources and resulting in edge server overload. This paper proposes a service migration method for resource competition, which can eliminate the problem of edge server overload caused by blind resource competition among users. Experimental results show that, compared with the existing algorithms, proposed method has lower cost and higher availability in multi-user scenarios.