Aiming at the security threats and single point of failure of the agent in mimic cloud systems， a high-available mimic cloud agent with active-standby monitoring was proposed. Firstly， an active-standby monitoring mechanism based on distributed agents in the cloud environment was proposed to construct heterogeneous active-standby agents. The traffic to the active agent was analyzed by the standby agent through mirroring the traffic， and the output results of the active agent were cross-validated by the standby agent. Secondly， based on the Data Plane Development Kit （DPDK） platform， a synchronous adjudication mechanism for standby agents and a seamless active-standby switching mechanism were designed to achieve security reinforcement and performance optimization of cloud agents. Finally， an active-standby switching decision algorithm was proposed to avoid the waste of resources caused by frequent active/standby switching. Experimental results showed that compared with the traffic processing delay of Nginx based cloud agents， the loss of this mimic cloud agent was milliseconds under high concurrency. It can be seen that the designed method can greatly improve the security and stability of the cloud proxy， and reduce the impact of the single point of failure on the stability of the proxy.