Extended Features of Network File Systems

Several key features of network file systems differentiate them from single-system file systems, as discussed in the following sections.

Disk Quotas

Considering that network file systems are a shared resource, it makes sense to control how it is used. Disk quotas set capacity limits for the total amount of storage individual users can consume. This can be done by associating a directory in the file system with a particular user and then limiting the size of that directory, including subdirectories.

Name Space Virtualization

The name space (directory and file organization) viewed through a client's I/O redirector can appear to be much different from the actual name space of the file server's file systems. For example, a client user could see several different files from multiple file systems as existing within the same virtual directory.

For the purposes of this book, we'll refer to the transposition of name space elements as name space virtualization. This technique is sometimes referred to as file-level virtualization, but it is probably best to specify that it is a name space manipulation and does not impact any of the underlying file system operations.

Name space virtualization can be accomplished at either the client side, the server side, or both. For instance, a server system could present the contents of several different directories as a single directory for clients, and each of those clients could aggregate files from several different server virtual directories. There are many ways file system views can be assembled.

Large Block Size Definitions

The file systems used for network file systems can be designed with large block sizes to fit certain applications. For instance, a network file system could have a block size definition of 2 MB for working with large engineering or multimedia data files. Large block sizes like this provide optimal throughput for streaming I/O applications.

Volume Management

Volume management software is an excellent technology match for network file systems. The redundancy, scalability, and performance benefits of volume management can be used very effectively in support of many client systems.

Basic volume management software is sometimes sold as part of an operating system, as it is with many Linux distributions. On Windows systems, however, volume management functionality is sold as part of Windows network file server products.

Storage Resource Management

File server software products usually include advanced management, diagnostics, and reporting capabilities to help administrators identify trends and problems. Storage resource management (SRM) is a set of functions that help administrators manage usage-related elements of storage.

For instance, SRM analyzes a file system's contents to determine the trends and dynamics of how its capacity is being consumed. For instance, an SRM report could identify files that are both greater than 5 MB and have not been accessed within the last 180 days. Obviously, this kind of analysis would be very helpful for administrators who need to identify ways to reclaim disk space when they are running low.

NOTE

There is no way to know how full a storage address space (volume) is without querying the file system. The file system is the only logical entity with information about how the file system has been used and how much free space there is.

Domain Services

File server systems often provide network domain services that help administrators structure network access and help clients locate and identify network resources, including network file system files and directories.

SAMBA

In 1992, a computer scientist named Andrew Tridgell at Australia National University in Canberra, Australia reverse-engineered the protocol used by Digital Equipment Corporation (DEC) for its Pathworks network file system product. In the years since then, Tridgell's software creation has evolved, expanded, and become an open-source software initiative called SAMBA (a filling out of the acronym SMB) for turning Linux systems into CIFS servers for Windows clients.

SAMBA is a CIFS file service running on Linux systems that allows Windows clients to store data on Linux-based file systems. It is possible to use any number of file systems available for Linux, but implementers need to pay attention to Linux system details (as is true for any open-source system software) to ensure implementations will work as planned.

As Microsoft continues to modify CIFS, the open-source developers working on SAMBA continue to reverse-engineer it and implement it in SAMBA. In addition to providing file services, SAMBA also provides Windows Server domain controller functionality. Readers interested in reading about SAMBA can go the organization's website at http://www.samba.org.