Summary

This chapter introduced a few more advanced topics in file systems used in network storage: cluster file systems, distributed file systems, and database systems.

Cluster file systems have to support the application and failover requirements of cluster operations. Two basic storage architectures are used with clusters: shared everything and shared nothing. Shared everything clusters provide the best performance and scalability, but shared nothing cluster storage is much simpler for the purposes of locking and memory mirroring.

Distributed file systems were then discussed as a way to implement the basic concepts of network file systems, but in a server farm environment. The chapter closed with a brief survey of the relationships between database systems and network storage technology. Several options are available for database storage controlled by a distributed file system.