High-Availability(HA) is a critical capacity in core router systems. The redundant backup is a widely used approach to achieve HA. It?s a big problem to backup the large amount of database in RAM associated with the running information in the router system. In this paper, we have studied the requirement of the memory backup procedure in core routers. Then, the cost of the conventional management is analyzed, and a new schema - Packed, Self-adaptive, Selfcontained scheme is proposed, which provide a new style of management policy for large amount of memory in routers. We also study the performance improvement by introducing the new approach. As a result, the PS^2 scheme can improve the backup performance greatly.