This paper describes an agent-based approach to the automatic provision of permanent virtual channels (PVC) in ATM networks. The agent framework described fosters flexibility and seamless evolution by the definition of capability skills. Skills are plugged into the agent's brain which is responsible for the coordination of the agent's behavior. The application of this agent framework to the problem of PVC configuration resulted in the definition of a set of skills, among which, only one skill is device-dependent. Therefore, the implemented application can easily support new types of ATM switches. Moreover, it is efficient in terms of performance and bandwidth saving.

We propose the introduction of a polling layer that allows the efficient instrumentation of management information models for the use of monitoring-intensive management applications. The polling layer allows the optimization of the number of polling packets sent to the SNMP agents to collect management data. We describe the architecture of the polling layer that allows the management applications to specify their exact requirements for polling operations. We show how the knowledge of these requirements, combined with an efficient use of SNMP packets, leads to an important gain in the generated polling packets. This gain gets more important under heavy monitoring load. Finally, we show the advantages of the polling layer over different other approaches and we discuss its tradeoffs.

Increasingly nowadays, networks are managed in a hierarchical, yet evolving to a distributed manner. The managed network is divided into sub-networks or domains that are managed more or less independently by autonomous agents. Once the failure of an agent is detected, it becomes even possible to have a further improvement by re-affecting the management tasks of the unreliable agent among the other agents in a way to ensure that the whole network continues to be reliably managed. This provides the property of graceful degradation to the distributed management system. The work presented in this paper provides a first step towards this interesting improvement. To ensure that the whole network is still managed even if a number of agents become unreliable, it is necessary to install a mechanism that continuously checks the reliability of the agents. When unreliable agents are detected, the management tasks that they have been performing are re-distributed amongst the other still-reliable agents. At some time in the future, the agent with the abnormal behavior might recover, for example following a human intervention, and the tasks that have been re-distributed on the other agents should be assigned back to the recovered agents.

We present an architecture for flexible intelligent agents dedicated to building network management systems and their application for a concrete experiment. The agent architecture is based on skills that can be plugged into an agent brain. The skill is a piece of code that can be loaded seamlessly into an agent to supply it with more new capabilities. The brain is a common part that governs the whole function of the agent. We will explain in detail how this architecture was used to build a fault-tolerant agent system that distributedly monitors a multi-domain network. The fault tolerance was ensured by a detection mechanism based on system level diagnosis, and a recovery procedure based on task redistribution. The experiment demonstrated how the agent system could ensure the distributed monitoring even in the case of the partial failure of the agent system.

Managing networks and distributed systems becomes every day harder than ever, because this activity is being simultaneously influenced by two distinct phenomena. First, the number of services offered is growing, ranging from more or less traditional data services (e.g. distributed file systems or intranet webs) to multimedia services (e.g. Video on Demand, Collaborative Work, Tele-Conferencing, etc). Second, and specially for multimedia services, users have now a set of complex requirements that they expect to be satisfied. In this paper we propose a framework towards an integration of user requirements into Network Management infrastructures. Therefore a dual role quality of service (QoS) framework is proposed, that allow applications to specify their QoS requirements, as well as allow management applications to handle those specifications so the appropriate management operations could be taken. The paper ends with the presentation of the management application principles and architectures suited for the proposed framework.

Current TMN-based information model standards take too long to specify, standardise, implement, test and introduce. The TMN-based Information Model Simulator-TIMS-explores the possibilities of reducing this development life cycle. The TIMS-vision foresees a formally executable specification of the behavior of object interfaces which should enable: (1) rapid-prototyping of TMN agent and manager functions and later allow for (2) the generation of function reference configurations as well as (3) the automatic derivation of test-suites based on the formal specifications.

The main purpose of this paper is to evaluate the performance of the HDLC (high-level data link control) protocol (to be more precise, we only consider the asynchronous response mode), and to suggest and analyze some possible improvements. The performance criteria are maximum throughput and response time and the flow of data is bidirectional with nonuniform frame sizes. The mathematical model of the HDLC protocol developed here allows us to compute its performance as a function of the maximum window width according to the characteristics of the traffic and taking into account the transmission delays and the overheads.

In this paper a new method for analyzing complex queueing networks is proposed: the isolation method. As an example, we study packet switching networks with finite buffer size at each node.

It IS shown that certain problems of optimal preemptive scheduling of unrelated parallel processors can be formulated and solved as hnear programming problems As a by-product of the linear programming formulaUons of these problems, upper bounds are obtained on the number of preempuons required for optimal schedules In particular it is shown that no more than O(m 2) preemptions are necessary, m order to schedule n jobs on m unrelated processors so as to minimize makespan

The purpose of this paper is to analyze and optimize the behavior of the broadcast channel for a packet transmission operating in the slotted mode Mathematical methods of Markov chain theory are used to prove the inherent lnstablhty of the system If no control is apphed, the effective throughput of the system will tend to zero tf the population of user terminals ~s sufficiently large Two classes of control pohcles are examined, the first acts on admissions to the channel from active terminals, and the second modifies the retransmlss~on rate of packets In each case sufflc~ent conditions for channel stability are given.In the case of retransm~sslon controls it is shown that only pohcles which assure a rate of retransmlsslon from each blocked terminal of the form off = 1/n, where n is the total number of blocked terminals, will yield a stable channel It ts also proved that the optimal pohcy which maximizes the maximum achievable throughput wtth a stable channel IS of the formf = (1 -k)/n Simulations illustrating channel lnstabdlty and the effect of the opnmal control are prowded KEY WORDS AND PHRASES computer networks, packet swnchlng, broadcast channel, optimal control CR CATEGORIES 4 3, 4 6, 5 5, 6 2, 6 35 ~t has to repeat the transmission untd it succeeds Either a termmal which is not blocked

Numérisation avec OCR réalisée en 2024. La reconnaissance de caractères du PDF (format PDF/A) peut comporter des erreurs. Pour toutes informations complémentaires et les partages de propriété, merci de contacter le service IES [email protected]