Handbook of Local Area Networks, 1998 Edition:Applications of LAN Technology Click Here! Search the site:   ITLibrary ITKnowledge EXPERT SEARCH Programming Languages Databases Security Web Services Network Services Middleware Components Operating Systems User Interfaces Groupware & Collaboration Content Management Productivity Applications Hardware Fun & Games EarthWeb sites Crossnodes Datamation Developer.com DICE EarthWeb.com EarthWeb Direct ERP Hub Gamelan GoCertify.com HTMLGoodies Intranet Journal IT Knowledge IT Library JavaGoodies JARS JavaScripts.com open source IT RoadCoders Y2K Info Previous Table of Contents Next Operating System: UNIX versus NT When choosing a server, the operating system deserves particular attention. The choices are usually between UNIX and Windows NT. Although some vendors offer server software for Windows 3.1 and Windows 95, these are usually intended for casual rather than business use. Most Internet servers are based on UNIX, but Windows NT is growing in popularity and may overtake UNIX in the near future. A Windows NT server offers comparable performance and functionality to a UNIX server and is easier to set up and administer, making it the platform of choice among developers of new sites. Like UNIX, Windows NT is a multitasking, multithreaded operating system. As such, NT executes software as threads, which are streams of commands that make up applications. At any point during execution, NT’s process manager interrupts (or preempts) a thread to allow another thread some CPU time. Also like UNIX, Windows NT supports multiple processors. If the server has more than one CPU, NT distributes the threads over the processors, allowing two or more threads to run simultaneously. Fault Tolerance If the server is supporting mission-critical applications over the Internet, several levels of fault tolerance merit consideration. Fault tolerance must be viewed from both the systems and subsystems perspectives. Site Mirroring From the systems perspective, fault tolerance can be implemented by linking multiple servers together. When one system fails or must be taken offline for upgrades or reconfigurations, the standby system is activated to handle the load. This is often called site mirroring. An additional level of protection can be obtained through features of the operating system that protect read and write processes in progress during the switch to the standby system. Hot Standby At the subsystem level, there are several server options that can improve fault tolerance, including ports, network interfaces, memory expansion cards, disks, tapes, and I/O channels. All must be duplicated so that an alternate hardware component can assume responsibility in the event of a subsystem failure. This procedure is sometimes referred to as a hot-standby solution, whereby a secondary subsystem monitors the tasks of the primary subsystem in preparation for assuming such tasks when needed. If a component in the primary subsystem fails, the secondary subsystem takes over without users being aware that a changeover has taken place. An obvious disadvantage of this solution is that companies must purchase twice the amount of hardware needed, and half of this hardware remains idle unless a failure occurs in the primary system. Because large amounts of 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 be able to detect an incomplete transaction so that the data base is not left in an inconsistent state. The server’s rollback facility is invoked automatically, which backs out of the partially updated data base. 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. Load Balancing Another means of achieving fault tolerance is to have all hardware components function simultaneously, but with a load-balancing mechanism that reallocates the processing tasks to surviving components when a failure occurs. This technique requires a UNIX operating system equipped with vendor options that continually monitor the system for errors and dynamically reconfigures the system to adapt to performance problems. Hot Swapping Hot swapping is an important capability that allows the network administrator to remove and replace faulty server modules without interrupting or degrading network performance. In some cases, standby modules can be brought online through commands issued at the network management workstation or automatically upon fault detection. Uninterruptible Power Supply To guard against an onsite power outage, an uninterruptible power supply (UPS) can provide an extra measure of protection. The UPS provides enough standby power to permit continuous operation or an orderly shutdown during power failures, or to change over to other power sources such as diesel-powered generators. Some UPSs have simple network management protocol (SNMP) capabilities, so network managers can monitor battery backup from the central management console. For example, using SNMP, every UPS can be instructed to test itself once a week and report back if the test fails. INTERNET APPLICATION SOFTWARE An Internet server must be equipped with software that allows it to run various Internet applications. Some server software supports general communications for document publishing over the World Wide Web. Often called a communications server or Web server, this type of server can be enhanced with software specifically designed for secure electronic commerce. Server software is available for performing many different functions, including implementing newsgroups, facilitating message exchange (i.e., E-mail), improving the performance and security of communications, and controlling traffic between the Internet and the corporate network. Sometimes a server is dedicated to a single application such as E-mail, newsgroups, or electronic commerce. Other times, the server supports multiple Internet applications. The specific configuration depends on such factors as available system resources (i.e., memory, disk space, processing power, and port capacity), network topology, available bandwidth, traffic patterns, and the security requirements of the organization. Previous Table of Contents Next Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.