Packet Transport Network Recovery System based on Failure Pattern  
Author Toshiaki Suzuki


Co-Author(s) Hiroyuki Kubo; Hayato Hoshihara; Taro Ogawa


Abstract A packet transport network recovery system that is able to recover from not only a single failure of a node but also catastrophic failures is proposed. A network is separated into multiple areas and failures are segmented into three patterns: single failure of a node, failures of multiple nodes, and failures of multiple network areas. The single failure is recovered by a protection scheme. Backup paths for the failures of multiple nodes are prepared for each combination of up to K nodes. If there are failures of over K nodes, they are regarded as failures of network areas and backup paths are prepared for entire combinations of multiple areas. A unique recovery ID is assigned to each failure pattern and backup paths with the recovery ID are stored in each node before network operations start. When network failures occur, the network management server determines the type of failure and sends the appropriate recovery ID to the nodes. Then recovery paths are configured. A prototype system composed of 100 emulated nodes was developed. Our proposed system took only about 0.5 seconds to configure all 1,000 backup paths after failures were detected, compared to about four seconds by a conventional restoration scheme.


Keywords packet transport network, network management, backup operation plane, failure recovery
    Article #:  24242
Proceedings ISSAT International Conference on Reliability and Quality in Design 2018
August 2-4, 2018 - Toronto, Ontario, Canada