Thesis (M.S.)--University of Texas at El Paso, 2008. / Title from title screen. Vita. CD-ROM. Includes bibliographical references. Also available online.

Thesis (MScEng)--University of Stellenbosch, 2003. / ENGLISH ABSTRACT: The Linux operating system and shared Ethernet are alternative technologies with the potential to reduce both the development time and costs of satellites as well as the supporting infrastructure. Modular satellites, ground stations and rapid proto typing testbeds also have a common requirement for distributed real-time computation. The identified technologies were investigated to determine whether this requirement could also be met. Various real-time extensions and modifications are currently available for the Linux operating system. A suitable open source real-time extension called Real-Time Application Interface (RTAI) was selected for the implementation of an experimental distributed real-time system. Experimental results showed that the RTAI operating system could deliver deterministic realtime performance, but only in the absence of non-real-time load. Shared Ethernet is currently the most popular and widely used commercial networking technology. However, Ethernet wasn't developed to provide real-time performance. Several methods have been proposed in literature to modify Ethernet for real-time communications. A token passing protocol was found to be an effective and least intrusive solution. The Real-Time Token (RTToken) protocol was designed to guarantee predictable network access to communicating real-time tasks. The protocol passes a token between nodes in a predetermined order and nodes are assigned fixed token holding times. Experimental results proved that the protocol offered predictable network access with bounded jitter. An experimental distributed real-time system was implemented, which included the extension of the RTAI operating system with the RTToken protocol, as a loadable kernel module. Real-time tasks communicated using connectionless Internet protocols. The Real-Time networking (RTnet) subsystem of RTAI supported these protocols. Under collision-free conditions consistent transmission delays with bounded jitter was measured. The integrated RTToken protocol provided guaranteed and bounded network access to communicating real-time tasks, with limit overheads. Tests exhibited errors in some of the RTAI functionality. Overall the investigated technologies showed promise in being able to meet the distributed real-time requirements of various applications, including those found in the satellite environment. / AFRIKAANSE OPSOMMING: Die Linux bedryfstelsel en gedeelde Ethernet is geïdentifiseer as potensiële tegnologieë vir satelliet bedryf wat besparings in koste en vinniger ontwikkeling te weeg kan bring. Modulêr ontwerpte satelliete, grondstasies en ontwikkeling platforms het 'n gemeenskaplike behoefte vir verspreide intydse verwerking. Verskillende tegnologieë is ondersoek om te bepaal of aan die vereiste ook voldoen kan word. Verskeie intydse uitbreidings en modifikasies is huidiglik beskikbaar vir die Linux bedryfstelsel. Die "Real-Time Application Interface" (RTAI) bedryfstelsel is geïdentifiseer as 'n geskikte intydse uitbreiding vir die implementering van 'n eksperimentele verspreide intydse stelsel. Eksperimentele resultate het getoon dat die RTAI bedryfstelsel deterministies en intyds kan opereer, maar dan moet dit geskied in die afwesigheid van 'n nie-intydse verwerkingslas. Gedeelde Ethernet is 'n kommersiële network tegnologie wat tans algemeen beskikbaar is. Die tegnologie is egter nie ontwerp vir intydse uitvoering nie. Verskeie metodes is in die literatuur voorgestelom Ethernet te modifiseer vir intydse kommunikasie. Hierdie ondersoek het getoon dat 'n teken-aangee protokol die mees effektiewe oplossing is en waarvan die implementering min inbreuk maak. Die "Real-Time Token" (RTToken) protokol is ontwerp om voorspelbare netwerk toegang tot kommunikerende intydse take te verseker. Die protokol stuur 'n teken tussen nodusse in 'n voorafbepaalde volgorde. Nodusse word ook vaste teken hou-tye geallokeer. Eksperimentele resultate het aangedui dat die protokol deterministiese netwerk toegang kan verseker met begrensde variasies. 'n Eksperimentele verspreide intydse stelsel is geïmplementeer. Dit het ingesluit die uitbreiding van die RTAI bedryfstelsel met die RTToken protokol; verpak as 'n laaibare bedryfstelsel module. Intydse take kan kommunikeer met verbindinglose protokolle wat deur die "Real-Time networking" (RTnet) substelsel van RTAI ondersteun word. Onder ideale toestande is konstante transmissie vertragings met begrensde variasies gemeet. Die integrasie van die RTToken protokol het botsinglose netwerk toegang aan kommunikerende take verseker, met beperkte oorhoofse koste as teenprestasie. Eksperimente het enkele foute in die funksionaliteit van RTAI uitgewys. In die algemeen het die voorgestelde tegnologieë getoon dat dit potensiaal het vir verskeie verspreide intydse toepassings in toekomstige satelliet en ook ander omgewings.

The World Wide Web is becoming a basic infrastructure for a variety of services, and the increases in audience size and client network bandwidth create service demands that are outpacing server capacity. Web clusters are one solution to this need for highperformance, highly available web server systems. We are interested in load distribution techniques, specifically Layer-7 algorithms that are content-aware. Layer- 7 algorithms allow distribution control based on the specific content requested, which is advantageous for a system that offers highly heterogenous services. We examine the performance of the Client Aware Policy (CAP) on a Linux/Apache web cluster consisting of a single web switch that directs requests to a pool of dual-processor SMP nodes. We show that the performance advantage of CAP over simple algorithms such as random and round-robin is as high as 29% on our testbed consisting of a mixture of static and dynamic content. Under heavily loaded conditions however, the performance decreases to the level of random distribution. In studying SMP vs. uniprocessor performance using the same number of processors with CAP distribution, we find that SMP dual-processor nodes under moderate workload levels provide equivalent throughput as the same number of CPU’s in a uniprocessor cluster. As workload increases to a heavily loaded state however, the SMP cluster shows reduced throughput compared to a cluster using uniprocessor nodes. We show that the web cluster’s maximum throughput increases linearly with the addition of more nodes to the server pool. We conclude that CAP is advantageous over random or round-robin distribution under certain conditions for highly dynamic workloads, and suggest some future enhancements that may improve its performance.