Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software

The NFS protocol continues to evolve and develop with strong support from the industry. One example of this is the continuing effort that was started with NFSv4^[26] file delegation to remove the NFS server performance bottleneck by distributing file I/O operations to NFS clients. With NFSv4, delegation of the applications running on an NFS client can lock different byte-range portions of the file without creating any additional network traffic to the NAS server. This has led to a development effort called NFS Extensions for Parallel Storage. The goal of this effort is to build a highly available filesystem with no single point of failure by allowing NFS clients to share file state information and file data without requiring communication with a single NAS device for each filesystem operation.

For more information about the NFS Extensions for Parallel Storage, see http://www.citi.umich.edu/NEPS.

^[26]NFSv4 (see RFC 3530) will provide a much better security model and allow for better interoperability with CIFS. NFSv4 will also provide better handling of client crashes and robust request ordering semantics as well as a facility for mandatory file locking.