A Virtual Deadline Scheduler for Window-Constrained Service Guarantees

This paper presents a new approach to window-constrained scheduling, suitable for multimedia and weakly-hard real-time systems. We originally developed an algorithm, called Dynamic Window-Constrained Scheduling (DWCS), that attempts to guarantee no more than x out of y deadlines are missed for real-time jobs such as periodic CPU tasks, or delay-constrained packet streams. While DWCS is capable of generating a feasible window-constrained schedule that utilizes 100% of resources, it requires all jobs to have the same request periods (or intervals between successive service requests). We describe a new algorithm called Virtual Deadline Scheduling (VDS), that provides window-constrained service guarantees to jobs with potentially different request periods, while still maximizing resource utilization. VDS attempts to service m out of k job instances by their virtual deadlines, that may be some finite time after the corresponding real-time deadlines. Notwithstanding, VDS is capable of outperforming DWCS and similar algorithms, when servicing jobs with potentially different request periods. Additionally, VDS is able to limit the extent to which a fraction of all job instances are serviced late. Results from simulations show that VDS can provide better window-constrained service guarantees than other related algorithms, while still having as good or better delay bounds for all scheduled jobs. Finally, an implementation of VDS in the Linux kernel compares favorably against DWCS for a range of scheduling loads.

Service guarantees and customer satisfaction: an empirical study of guarantee type and service experience

Scenario analysis was used to examine empirically the relationships between guarantee type and service experience, and consumer satisfaction, for the service of an Internet Service Provider (ISP). The scenarios involved hypothetical situations in which several factors were varied: the existence of a problem; the invocation of a guarantee, the identity of the invoker; and the manner of resolution of any problem. Alternative service guarantees were associated with each hypothetical experience: a specific guarantee, and an unconditional guarantee. Overall, consumer satisfaction related to the nature of the service experience much more strongly than it did to the difference in guarantee type.

Service failure and recovery in the presence of service guarantees

This paper presents a conceptual model and propositions outlining how, in a service recovery encounter, service guarantees (unconditional and specific) operate in conjunction with other organisational recovery variables (guarantee facilitation and service provider concern), to influence customers’ justice perceptions and subsequent satisfaction evaluations.

There's a fly in my soup : the influence of service guarantees and personal requests on customer voice

Anecdotal evidence suggests that service guarantees and personal requests by service workers encourage customers to voice following failure. However, empirical support for these tactics in facilitating complaints to the organisation is limited. To address this deficiency, a 3 (guarantee treatment: none, unconditional or combined) x 2 (personal request to voice: yes or no) x 2 (failure severity: minor or major) full factorial, between subjects experiment was conducted in a restaurant context. Findings suggest that offering a service guarantee, regardless of whether it is unconditional or combined, can encourage voice. Severity of the failure was also found to be associated with voice. Surprisingly, however, a personal request to voice was not related to customers’ voice intentions. Implications of the findings are discussed.

Service guarantee redress options in a hotel environment

Unconditional service guarantees are a popular marketing tool in the hotel industry worldwide. They promise total satisfaction and guests are free to invoke the guarantee whenever they are dissatisfied. While many hotels offer “money-back” compensation following guarantee invocation, others vary the payout depending on the severity of the service failure and still others will only compensate the customer if the problem leading to invocation of the guarantee cannot be fixed. To the researcher’s knowledge, the influence of compensation and fix (i.e., taking action to resolve the problem) on consumers’ perceptions of distributive justice has not been examined previously in a service guarantee context. This paper begins to address this gap by presenting a conceptual model and related propositions, arguing that redress (compensation and fix) is an important predictor of consumers’ perceptions of distributive justice, and that this relationship is moderated by service failure severity.

The influence of failure severity and perceived employee effort on consumers' postcomplaint negative WOM intentions in a service guarantee context

This study employed a 2 x 2 full-factorial, between-subjects design experiment examining the influence of failure severity and perceived employee effort on hotel guests’ negative word-ofmouth (WOM) intentions following invocation of a service guarantee. The study involved a sample of 131 online panel members. Results suggest that negative WOM intentions reduced when a greater level of effort is exerted by staff in rectifying the guest’s problem and increased when a more severe failure is experienced. There is a stronger difference in guests’ negative WOM intentions between the high and low employee effort conditions when a minor versus a severe service failure is experienced by guests.

Quality of Service (QoS) in MPLS networks

Telecommunications networks have been always expanding and thanks to it, new services have appeared. The old mechanisms for carrying packets have become obsolete due to the new service requirements, which have begun working in real time. Real time traffic requires strict service guarantees. When this traffic is sent through the network, enough resources must be given in order to avoid delays and information losses. When browsing through the Internet and requesting web pages, data must be sent from a server to the user. If during the transmission there is any packet drop, the packet is sent again. For the end user, it does not matter if the webpage loads in one or two seconds more. But if the user is maintaining a conversation with a VoIP program, such as Skype, one or two seconds of delay in the conversation may be catastrophic, and none of them can understand the other. In order to provide support for this new services, the networks have to evolve. For this purpose MPLS and QoS were developed. MPLS is a packet carrying mechanism used in high performance telecommunication networks which directs and carries data using pre-established paths. Now, packets are forwarded on the basis of labels, making this process faster than routing the packets with the IP addresses. MPLS also supports Traffic Engineering (TE). This refers to the process of selecting the best paths for data traffic in order to balance the traffic load between the different links. In a network with multiple paths, routing algorithms calculate the shortest one, and most of the times all traffic is directed through it, causing overload and packet drops, without distributing the packets in the other paths that the network offers and do not have any traffic. But this is not enough in order to provide the real time traffic the guarantees it needs. In fact, those mechanisms improve the network, but they do not make changes in how the traffic is treated. That is why Quality of Service (QoS) was developed. Quality of service is the ability to provide different priority to different applications, users, or data flows, or to guarantee a certain level of performance to a data flow. Traffic is distributed into different classes and each of them is treated differently, according to its Service Level Agreement (SLA). Traffic with the highest priority will have the preference over lower classes, but this does not mean it will monopolize all the resources. In order to achieve this goal, a set policies are defined to control and alter how the traffic flows. Possibilities are endless, and it depends in how the network must be structured. By using those mechanisms it is possible to provide the necessary guarantees to the real-time traffic, distributing it between categories inside the network and offering the best service for both real time data and non real time data. Las Redes de Telecomunicaciones siempre han estado en expansión y han propiciado la aparición de nuevos servicios. Los viejos mecanismos para transportar paquetes se han quedado obsoletos debido a las exigencias de los nuevos servicios, que han comenzado a operar en tiempo real. El tráfico en tiempo real requiere de unas estrictas garantías de servicio. Cuando este tráfico se envía a través de la red, necesita disponer de suficientes recursos para evitar retrasos y pérdidas de información. Cuando se navega por la red y se solicitan páginas web, los datos viajan desde un servidor hasta el usuario. Si durante la transmisión se pierde algún paquete, éste se vuelve a mandar de nuevo. Para el usuario final, no importa si la página tarda uno o dos segundos más en cargar. Ahora bien, si el usuario está manteniendo una conversación usando algún programa de VoIP (como por ejemplo Skype) uno o dos segundos de retardo en la conversación podrían ser catastróficos, y ninguno de los interlocutores sería capaz de entender al otro. Para poder dar soporte a estos nuevos servicios, las redes deben evolucionar. Para este propósito se han concebido MPLS y QoS MPLS es un mecanismo de transporte de paquetes que se usa en redes de telecomunicaciones de alto rendimiento que dirige y transporta los datos de acuerdo a caminos preestablecidos. Ahora los paquetes se encaminan en función de unas etiquetas, lo cual hace que sea mucho más rápido que encaminar los paquetes usando las direcciones IP. MPLS también soporta Ingeniería de Tráfico (TE). Consiste en seleccionar los mejores caminos para el tráfico de datos con el objetivo de balancear la carga entre los diferentes enlaces. En una red con múltiples caminos, los algoritmos de enrutamiento actuales calculan el camino más corto, y muchas veces el tráfico se dirige sólo por éste, saturando el canal, mientras que otras rutas se quedan completamente desocupadas. Ahora bien, esto no es suficiente para ofrecer al tráfico en tiempo real las garantías que necesita. De hecho, estos mecanismos mejoran la red, pero no realizan cambios a la hora de tratar el tráfico. Por esto es por lo que se ha desarrollado el concepto de Calidad de Servicio (QoS). La calidad de servicio es la capacidad para ofrecer diferentes prioridades a las diferentes aplicaciones, usuarios o flujos de datos, y para garantizar un cierto nivel de rendimiento en un flujo de datos. El tráfico se distribuye en diferentes clases y cada una de ellas se trata de forma diferente, de acuerdo a las especificaciones que se indiquen en su Contrato de Tráfico (SLA). EL tráfico con mayor prioridad tendrá preferencia sobre el resto, pero esto no significa que acapare la totalidad de los recursos. Para poder alcanzar estos objetivos se definen una serie de políticas para controlar y alterar el comportamiento del tráfico. Las posibilidades son inmensas dependiendo de cómo se quiera estructurar la red. Usando estos mecanismos se pueden proporcionar las garantías necesarias al tráfico en tiempo real, distribuyéndolo en categorías dentro de la red y ofreciendo el mejor servicio posible tanto a los datos en tiempo real como a los que no lo son.

Unicast Routing: Cost-Performance Tradeoffs

The objective of unicast routing is to find a path from a source to a destination. Conventional routing has been used mainly to provide connectivity. It lacks the ability to provide any kind of service guarantees and smart usage of network resources. Improving performance is possible by being aware of both traffic characteristics and current available resources. This paper surveys a range of routing solutions, which can be categorized depending on the degree of the awareness of the algorithm: (1) QoS/Constraint-based routing solutions are aware of traffic requirements of individual connection requests; (2) Traffic-aware routing solutions assume knowledge of the location of communicating ingress-egress pairs and possibly the traffic demands among them; (3) Routing solutions that are both QoS-aware as (1) and traffic-aware as (2); (4) Best-effort solutions are oblivious to both traffic and QoS requirements, but are adaptive only to current resource availability. The best performance can be achieved by having all possible knowledge so that while finding a path for an individual flow, one can make a smart choice among feasible paths to increase the chances of supporting future requests. However, this usually comes at the cost of increased complexity and decreased scalability. In this paper, we discuss such cost-performance tradeoffs by surveying proposed heuristic solutions and hybrid approaches.

Comparison of K-ary N-cube and de Bruijn Overlays in QoS-Constrained Multicast Applications

Research on the construction of logical overlay networks has gained significance in recent times. This is partly due to work on peer-to-peer (P2P) systems for locating and retrieving distributed data objects, and also scalable content distribution using end-system multicast techniques. However, there are emerging applications that require the real-time transport of data from various sources to potentially many thousands of subscribers, each having their own quality-of-service (QoS) constraints. This paper primarily focuses on the properties of two popular topologies found in interconnection networks, namely k-ary n-cubes and de Bruijn graphs. The regular structure of these graph topologies makes them easier to analyze and determine possible routes for real-time data than complete or irregular graphs. We show how these overlay topologies compare in their ability to deliver data according to the QoS constraints of many subscribers, each receiving data from specific publishing hosts. Comparisons are drawn on the ability of each topology to route data in the presence of dynamic system effects, due to end-hosts joining and departing the system. Finally, experimental results show the service guarantees and physical link stress resulting from efficient multicast trees constructed over both kinds of overlay networks.

Management of SPMD based parallel processing on clusters of workstations

Current attempts to manage parallel applications on Clusters of Workstations (COWs) have either generally followed the parallel execution environment approach or been extensions to existing network operating systems, both of which do not provide complete or satisfactory solutions. The efficient and transparent management of parallelism within the COW environment requires enhanced methods of process instantiation, mapping of parallel process to workstations, maintenance of process relationships, process communication facilities, and process coordination mechanisms. The aim of this research is to synthesise, design, develop and experimentally study a system capable of efficiently and transparently managing SPMD parallelism on a COW. This system should both improve the performance of SPMD based parallel programs and relieve the programmer from the involvement into parallelism management in order to allow them to concentrate on application programming. It is also the aim of this research to show that such a system, to achieve these objectives, is best achieved by adding new special services and exploiting the existing services of a client/server and microkernel based distributed operating system. To achieve these goals the research methods of the experimental computer science should be employed. In order to specify the scope of this project, this work investigated the issues related to parallel processing on COWs and surveyed a number of relevant systems including PVM, NOW and MOSIX. It was shown that although the MOSIX system provide a number of good services related to parallelism management, none of the system forms a complete solution. The problems identified with these systems include: instantiation services that are not suited to parallel processing; duplication of services between the parallelism management environment and the operating system; and poor levels of transparency. A high performance and transparent system capable of managing the execution of SPMD parallel applications was synthesised and the specific services of process instantiation, process mapping and process interaction detailed. The process instantiation service designed here provides the capability to instantiate parallel processes using either creation or duplication methods and also supports multiple and group based instantiation which is specifically design for SPMD parallel processing. The process mapping service provides the combination of process allocation and dynamic load balancing to ensure the load of a COW remains balanced not only at the time a parallel program is initialised but also during the execution of the program. The process interaction service guarantees to maintain transparently process relationships, communications and coordination services between parallel processes regardless of their location within the COW. The combination of these services provides an original architecture and organisation of a system that is capable of fully managing the execution of SPMD parallel applications on a COW. A logical design of a parallelism management system was developed derived from the synthesised system and was shown that it should ideally be based on a distributed operating system employing the client server model. The client/server based distributed operating system provides the level of transparency, modularity and flexibility necessary for a complete parallelism management system. The services identified in the synthesised system have been mapped to a set of server processes including: Process Instantiation Server providing advanced multiple and group based process creation and duplication; Process Mapping Server combining load collection, process allocation and dynamic load balancing services; and Process Interaction Server providing transparent interprocess communication and coordination. A Process Migration Server was also identified as vital to support both the instantiation and mapping servers. The RHODOS client/server and microkernel based distributed operating system was selected to carry out research into the detailed design and to be used for the implementation this parallelism management system. RHODOS was enhanced to provide the required servers and resulted in the development of the REX Manager, Global Scheduler and Process Migration Manager to provide the services of process instantiation, mapping and migration, respectively. The process interaction services were already provided within RHODOS and only required some extensions to the existing Process Manager and IPC Managers. Through a variety of experiments it was shown that when this system was used to support the execution of SPMD parallel applications the overall execution times were improved, especially when multiple and group based instantiation services are employed. The RHODOS PMS was also shown to greatly reduce the programming burden experienced by users when writing SPMD parallel applications by providing a small set of powerful primitives specially designed to support parallel processing. The system was also shown to be applicable and has been used in a variety of other research areas such as Distributed Shared Memory, Parallelising Compilers and assisting the port of PVM to the RHODOS system. The RHODOS Parallelism Management System (PMS) provides a unique and creative solution to the problem of transparently and efficiently controlling the execution of SPMD parallel applications on COWs. Combining advanced services such as multiple and group based process creation and duplication; combined process allocation and dynamic load balancing; and complete COW wide transparency produces a totally new system that addresses many of the problems not addressed in other systems.

Unificación de los protocolos de multipunto fiable optimizando la escalabilidad y el retardo

Las aplicaciones distribuidas que precisan de un servicio multipunto fiable son muy numerosas, y entre otras es posible citar las siguientes: bases de datos distribuidas, sistemas operativos distribuidos, sistemas de simulación interactiva distribuida y aplicaciones de distribución de software, publicaciones o noticias. Aunque en sus orígenes el dominio de aplicación de tales sistemas distribuidos estaba reducido a una única subred (por ejemplo una Red de Área Local) posteriormente ha surgido la necesidad de ampliar su aplicabilidad a interredes. La aproximación tradicional al problema del multipunto fiable en interredes se ha basado principalmente en los dos siguientes puntos: (1) proporcionar en un mismo protocolo muchas garantías de servicio (por ejemplo fiabilidad, atomicidad y ordenación) y a su vez algunas de éstas en distintos grados, sin tener en cuenta que muchas aplicaciones multipunto que precisan fiabilidad no necesitan otras garantías; y (2) extender al entorno multipunto las soluciones ya adoptadas en el entorno punto a punto sin considerar las características diferenciadoras; y de aquí, que se haya tratado de resolver el problema de la fiabilidad multipunto con protocolos extremo a extremo (protocolos de transporte) y utilizando esquemas de recuperación de errores, centralizados (las retransmisiones se hacen desde un único punto, normalmente la fuente) y globales (los paquetes solicitados se vuelven a enviar al grupo completo). En general, estos planteamientos han dado como resultado protocolos que son ineficientes en tiempo de ejecución, tienen problemas de escalabilidad, no hacen un uso óptimo de los recursos de red y no son adecuados para aplicaciones sensibles al retardo. En esta Tesis se investiga el problema de la fiabilidad multipunto en interredes operando en modo datagrama y se presenta una forma novedosa de enfocar el problema: es más óptimo resolver el problema de la fiabilidad multipunto a nivel de red y separar la fiabilidad de otras garantías de servicio, que pueden ser proporcionadas por un protocolo de nivel superior o por la propia aplicación. Siguiendo este nuevo enfoque se ha diseñado un protocolo multipunto fiable que opera a nivel de red (denominado RMNP). Las características más representativas del RMNP son las siguientes; (1) sigue una aproximación orientada al emisor, lo cual permite lograr un grado muy alto de fiabilidad; (2) plantea un esquema de recuperación de errores distribuido (las retransmisiones se hacen desde ciertos encaminadores intermedios que siempre estarán más cercanos a los miembros que la propia fuente) y de ámbito restringido (el alcance de las retransmisiones está restringido a un cierto número de miembros). Este esquema hace posible optimizar el retardo medio de distribución y disminuir la sobrecarga introducida por las retransmisiones; (3) incorpora en ciertos encaminadores funciones de agregación y filtrado de paquetes de control, que evitan problemas de implosión y reducen el tráfico que fluye hacia la fuente. Con el fin de evaluar el comportamiento del protocolo diseñado, se han realizado pruebas de simulación obteniéndose como principales conclusiones que, el RMNP escala correctamente con el tamaño del grupo, hace un uso óptimo de los recursos de red y es adecuado para aplicaciones sensibles al retardo.---ABSTRACT---There are many distributed applications that require a reliable multicast service, including: distributed databases, distributed operating systems, distributed interactive simulation systems and distribution applications of software, publications or news. Although the application domain of distributed systems of this type was originally confíned to a single subnetwork (for example, a Local Área Network), it later became necessary extend their applicability to internetworks. The traditional approach to the reliable multicast problem in internetworks is based mainly on the following two points: (1) provide a lot of service guarantees in one and the same protocol (for example, reliability, atomicity and ordering) and different levéis of guarantee in some cases, without taking into account that many multicast applications that require reliability do not need other guarantees, and (2) extend solutions adopted in the unicast environment to the multicast environment without taking into account their distinctive characteristics. So, the attempted solutions to the multicast reliability problem were end-to-end protocols (transport protocols) and centralized error recovery schemata (retransmissions made from a single point, normally the source) and global error retrieval schemata (the requested packets are retransmitted to the whole group). Generally, these approaches have resulted in protocols that are inefficient in execution time, have scaling problems, do not make optimum use of network resources and are not suitable for delay-sensitive applications. Here, the multicast reliability problem is investigated in internetworks operating in datagram mode and a new way of approaching the problem is presented: it is better to solve to the multicast reliability problem at network level and sepárate reliability from other service guarantees that can be supplied by a higher protocol or the application itself. A reliable multicast protocol that operates at network level (called RMNP) has been designed on the basis of this new approach. The most representative characteristics of the RMNP are as follows: (1) it takes a transmitter-oriented approach, which provides for a very high reliability level; (2) it provides for an error retrieval schema that is distributed (the retransmissions are made from given intermedíate routers that will always be closer to the members than the source itself) and of restricted scope (the scope of the retransmissions is confined to a given number of members), and this schema makes it possible to optimize the mean distribution delay and reduce the overload caused by retransmissions; (3) some routers include control packet aggregation and filtering functions that prevent implosión problems and reduce the traffic flowing towards the source. Simulation test have been performed in order to evalúate the behaviour of the protocol designed. The main conclusions are that the RMNP scales correctly with group size, makes optimum use of network resources and is suitable for delay-sensitive applications.