FINAL WORKSHOP OF GRID PROJECTS “PON RICERCA 2000-2006, AVVISO 1575” 1 A new metric to describe the efficiency of data communication networks. Elena Marchei 1 , Sandro Meloni 2 , Saverio Di Blasi 1 ,Vittorio Rosato 3,4 1 Ente per le Nuove Tecnologie, l’Energia e l’Ambiente (ENEA), Portici Research Center, Portici (Italy). 2 Dip.to Ingegneria Informatica, Università di Roma Tre, Via della Vasca Navale, Roma (Italy ). 3 Ente per le Nuove Tecnologie, l’Energia e l’Ambiente (ENEA), Casaccia Research Center, Roma (Italy) and 4 Ylichron Srl, Via Anguillarese 301, 00123 Roma (Italy). elena.marchei@enea.it, saverio.diblasi@enea.it, sandro.meloni@gmail.com , rosato@enea.it Abstract — This work has been focussed on the adaptive routing strategies able to prevent (or delay) the congestion transition. characterization of the congestion transition, occurring on model networks when traffic NETWORK’S TOPOLOGIES volume increases above a given threshold, and on its dependence on the network’s topology. We Two different network’s topologies (both have attempted to describe the transition using a formed by 1000 nodes and bi-directional links number of quantities which might be measured of equal bandwidth) have been generated; a on the different network’s components. We have random network, R-net, by using the Erdos- evaluated several quantities and showed how they behave above and below the congestion Renyi growth mechanism [1] and a Internet- threshold. In particular we were interested in like network, I-net, grown on a mechanism defining quantities which could provide each which couples Preferential Attachment [2] to network’s node a certain ”awareness” of the Triad Formation [3], as proposed in [4]. global state of the network. Synthetic Internet- like and Random networks have been generated Example of a R-net by using well-established growth mechanisms; data traffic behavior on the networks has been simulated by using either a simple traffic model (hereafter referred to as Traffic-Model) which reproduces the basic features of the traffic protocols and the NS2 tool which allows to simulate the whole TCP/IP stack. Data traffic undergoes a phase transition from a low-traffic phase of normal behaviour to a congested phase, characterized by a rapid increase on the value of the average delivery time of data packets and a sharp reduction of the Quality of Service, which accounts also for the number of effectively delivered data packets. We propose the use of a new metric to individuate the onset of the congestion transition. The proposed quantity, referred to as "node's temperature" allows to The Erdos-Renyi model is a random graph integrate, in an unique value, information on the generation model that produces graphs where local and the global state of data traffic. Node temperature can also be used to implement new the sticking probability of new nodes, in the growth mechanism, is independent on the node where the sticking is produced. In the resulting structure, nodes have approximately a similar
FINAL WORKSHOP OF GRID PROJECTS “PON RICERCA 2000-2006, AVVISO 1575” 1 degree and the degree distribution is a packet arrives at a node whose buffer is full, it poissonian distribution. is dropped and, in a first approximation, it is lost. Example of an I-net Simulations of data traffic have been performed by selecting a value λ measuring the traffic intensity: λ represents the frequency of emission of a new data packet by each node. λ = 0.1 will thus indicate that each node has a probability of 10% to generating, at each simulation time-step, a new data packet, directed towards a generic j node. NETWORK SIMULATOR NS2 AND ENEA-GRID NS2 is a discrete event simulator that together The Internet network has been established to with its companion Nam, the Network display a topology which is far from being Animator, form a very powerful set of tools for random. It belongs to the class of “scale-free” study of the real networks networks, where nodes with higher degree [www.isi.edu\nsnam\ns]. have stronger ability to grab links added to the NS2 allows to study e.g. wired and wireless network. In the growth process, nodes with (local and satellite) networks; create various higher degree have a larger probability to link routing algorithms; traffic sources like web, with new nodes. This mechanisms produces a ftp, telnet, cbr, stochastic traffic; failures, probability distribution of nodes P ( k ) ~ k - γ including deterministic, probabilistic loss, link where γ is a constant whose value is typically failure; various queuing disciplines. in the range 2< γ <3. To better mimic the When used in NS2, both R-net and I-net Internet topology, one must add to this models have been designed with a propagation mechanism a mechanisms called “triad delay of 30ms (which could be representative formation” which allows to generate scale-free of the times of signal processing) and queues networks with a high clustering coefficient type Drop-Tail, whose policy is to serve packets arriving with criterion FIFO (First In TRAFFIC-MODEL First Out) . Each node of the network represents a router Each node of the network is, at the same time, and it has a buffer of a limited size (in terms of both source and destination. number of packets which it can contain) able to The couples source-destination time are host data packets waiting to be delivered; to a generated in random way. We have used, as first approximation, buffers have equal size for transport protocol, an agent connectionless all the nodes of the network. UDP (User Datagram Protocol) and, as With a specific frequency, nodes send a small application traffic, a source with a rate constant data packet which flows along the network via CBR (Constant Bit Rate). Simulations of data the shortest path (a fixed-path routing is traffic have been performed by selecting a primarily established to rule the packets path). value of the source rate λ as packets per They are hosted in the node’s buffer, if buffer second . contains previously arrived packets which must To evaluate the network efficiency, we have be handled with priority, or dispatched toward used a unix script that allows us to launch to next node if the buffer nodes is empty. several simulations on different Linux hosts of Buffers, in fact, handle data packets with a ENEA-GRID. Each job runs simultaneously in FIFO (First-In-First-Out) policy. If a data
Recommend
More recommend
