In order to improve the availability of routers

systematically presented possible implementation approaches on high-availability routers and classified them into three catalogs: protocol extension

state synchronization and data dupli-cation.It modeled and analyzed these fault-recovery schemes in continuous-time Markov chains（CTMC）.Based on theo-retical results

it proposed a novel adaptive fault-recovery method（AFM） to dynamically select protection and restoration methods.It is illustrated that AFM could effectively improve the availability and is scalable.It has been implemented on a 128Gbit/s high-performance router.