3.5 Server Restoration Capabilities

---

Sharing resources distributed over the LAN can better protect users against the loss of information and unnecessary downtime than a network with all of its resources centralized at a single location. The vehicle for resource sharing is the server, which constitutes the heart of the LAN. The server gives the LAN its features, including those for security and data protection, as well as those for network management and resource accounting.

The server determines the friendliness of the user interface and governs the number of users that share the network at one time. It resides in one or more networking cards that are typically added to desktop or rack-mounted computers and which vary in processing power, memory, and interface capacity. However, servers are more than just hardware; they are also defined by the programs that provide services to clients. Aside from file servers and application servers, there are image and fax servers, e-mail servers, printer servers, structured query language (SQL) servers, and a variety of other specialized servers, including those for videoconferencing over the LAN.

The addition of multiple special-purpose servers provides the capability, connectivity, and processing power not provided by the network operating system and file server alone. A single multiprocessor server, combined with a network operating system designed to exploit its capabilities, such as UNIX, provides enough throughput to support five to ten times the number of users and applications as an ordinary PC that is used as a server. Bus and cache designs make it possible for the server to make full use of several processors at once, without the usual performance bottlenecks that slow application speed.

Distributing resources in this way minimizes the disruption to productivity that would result if all the resources were centralized and a failure were to occur.

Moreover, the use of specialized devices as servers permits the integration of diagnostic and maintenance capabilities not found in general-purpose PCs. Among these capabilities are error detection and correction, soft controller error detection and correction, and automatic shutdown in case of catastrophic error. Some servers include integral management functions (e.g., remote console management). The multiprocessing capabilities of specialized servers provide the power necessary to support the system overhead that all these sophisticated capabilities require.

Aside from physical faults on the network, there are various causes for erroneous-data. A software failure on the host, for example, can cause write errors to the user or server disk. Application software errors may generate bad values, or faults, on the disk itself. Power surges can corrupt data and application programs, while power outages can shut down sessions, wiping out data that has not yet been written to disk. Viruses and worms that are brought into the LAN from the Internet, external bulletin boards, shareware, and careless user uploads are another concern. User mistakes can also introduce errors into data or eliminate entire files. Although careful system administration and strict adherence to security procedures are usually sufficient to minimize most of these problems, they do not eliminate the need for backup and archival storage.

Many organizations follow traditional file backup procedures that can be implemented across the LAN. Some of these procedures include performing file backups at night—full backups if possible, incremental backups otherwise. Archival backups of all disk drives are typically done at least monthly; multiple daily saves of critical databases may be warranted in some cases. The more data users already have stored on their hard disks, the longer it takes to save. For this reason, LAN managers encourage users to off-load unneeded files and consolidate file fragments with utility software to conserve disk space, as well as to improve overall system performance during backups. Some LAN managers have installed automatic archiving facilities that will move files from users’ hard disks to a backup database if they have not been opened in the last 90 days.

As the amount of stored information increases, there is the need for LAN backup systems that address such strategic concerns as tape administration, disaster recovery, and the automatic movement of files up and down a hierarchy of network storage devices. Such capabilities are referred to as system storage management or hierarchical storage management.

Protecting data at the server has become a critical concern for most network managers; after all, a failure at the server can result in lost or destroyed data. Considering that some servers are capable of holding vast quantities of data in the terabyte range, loss or damage can have disastrous consequences for an information-intensive organization.

Because huge amounts of corporate data may be located at the server, the server must be able to implement recovery procedures in the event of a program, operating system, or hardware failure. For example, when a transaction terminates abnormally, the server must have the capability to detect an incomplete transaction so that the database is not left in an inconsistent state. The server’s rollback facility is invoked automatically, which backs out of the partially updated database. The transaction can then be resubmitted by the program or user. A roll-forward facility recovers completed transactions and updates in the event of a disk failure by reading a transaction journal that contains a record of all updates.

---