A File Server is a computer attached to a network that has the primary purpose of providing a location for the shared storage of information (such as documents, sound files, photographs, movies, images, databases, et cetera) that can be accessed by the workstations that are attached to the network. As distinguished from Timesharing or a Mainframe (where a central computer provides calculation resources for "dumb" terminals) a File Server is designed primarily to enable the rapid storage and retrieval of data where the heavy computation is provided by the workstations.
In the mid-1980's firms became increasingly interested in ways to connect a rapidly growing population of personal computers. Novell proposed an approach using software to connect each workstation to a network file server that would manage both the network and access to network resources. At the same time 3Com Corporation was creating Ethernet adapter cards building connectivity by way of a hardware approach.
Novell grew upon the strength of it's Netware and by the late 1980's had a 50% market share of Local Area Networks. Seeing the potential growth in this arena IBM and Microsoft Corp. both introduced their own file-server strategies.
Twenty years after its peak, Novell and 3Com have virtually disappeared from the Server market; today disk servers (a NAS or a SAN) supplied by one of the leading vendors such as EMC or Network Appliance are typically attached to a file server running the Microsoft Windows or Linux operating systems.
Large datacenters that serve millions of customers (such as Google) run server "farms" on specialized equipment with proprietary operating systems.
A file server may be dedicated or non-dedicated. A dedicated server is generally designed specifically for use as a file server, with workstations attached for reading and writing files and databases. A workstation may share files with other workstations on the network directly however by turning on its "Server" service and then by creating a "Share" -- this process creates a non-dedicated file server that is primarily being used as a workstation.
File servers may also be categorized by the method of access: remote servers are frequently accessed by FTP (they run an FTP Server) or by HTTP (they run a Web Server). Servers on a LAN are usually accessed by SMB/CIFS protocol http://http://en.wikipedia.org/wiki/Server_Message_Block (Windows, Linux, Unix and OS X) or NFS protocol http://en.wikipedia.org/wiki/Network_File_System_%28protocol%29 (Linux, Unix and OS X) protocols. Servers providing access to a shared database usually are accessed by a database device driver (they run a Database Server). Most servers provide access to printers that are attached across the network (they run a Print Spool). A single File Serving computer may be accessible by multiple means: it may run an FTP server, an HTTP server, a database server, and a print spool.
In modern businesses the design of file servers is complicated by competing demands for storage space, access speed, recoverability, ease of administration, compatibility, security, and budget. This is further complicated by a constantly changing design environment where new hardware and technology rapidly obsoletes old equipment and yet must come online in a fashion compatible with the older machinery. To manage throughput, peak loads, and response time, server manufacturers may utilize queuing theory to model how the combination of hardware and software will respond over various levels of demand. Servers may also employ dynamic Load Balancing to distribute requests across various pieces of hardware.
The primary piece of hardware equipment for servers over the last couple of decades has proven to be the disk drive. Although other forms of storage are viable (such as magnetic tape and RAM) disk drives have continued to offer the best fit for cost, performance, and capacity.
Since the crucial function of a file server is storage, hardware has been developed to operate multiple disk drives together as a team, forming a disk array. A disk array typically has cache (temporary memory storage that is faster than the magnetic disks) as well as the intelligence so that it can perform functions like RAID and virtualization. Typically disk arrays provide some level of redundancy by using mirrors and other redundant components. Disk arrays may be configured as a NAS, a SAN, or virtualized.
File servers generally offer some form of system security to limit access to files to specific users or groups. In large organizations, this is a task usually delegated to what is known as directory services such as openLDAP, Novell's eDirectory or Microsoft's Active Directory.
These servers work within the hierarchical computing environment which treat users, directories, computers, applications and files as distinct but related entities on the network and grant access based on user or group credentials. In many cases, the directory service spans many file servers, potentially hundreds for large organizations. In the past, and in smaller organizations, authentication can take place directly to the server itself.