A Highly Available Network File Server

This paper presents the design and implementation of a Highly Available Network File Server (HA-NFS). We separate the problem of network le server reliability into three diierent subprob-lems: server reliability, disk reliability, and network reliability. HA-NFS ooers a diierent solution for each: dual-ported disks and impersonation are used to provide server reliability, disk mirroring can be used to provide disk reliability, and optional network replication can be used to provide network reliability. The implementation shows that HA-NFS provides high availability without the excessive resource overhead or the performance degradation that characterize traditional replica-tion methods. Ongoing operations are not aborted during fail-over and recovery is completely transparent to applications. HA-NFS adheres to the NFS protocol standard and can be used by existing NFS clients without modiication.