1,721,083 research outputs found
Stochastic Analysis of Chain Based File Distribution Architectures with Heterogeneous Peers
No full textPerformance analysis of P2P systems is necessary to understand the real impact on the network of such applications. In this paper we study the performance that can be achieved by a simple file distribution architecture in a heterogeneous environment, i.e., when the access links of peers have randomly distributed capacities. The distribution architecture is a chain, where each peer downloads the content from exactly one node and uploads the content to exactly one node. Our analysis starts from a complete knowledge about peers, so that we can derive analytically the deterministic behavior and use the results as reference. We then remove part of the knowledge a peer has about its neighbors and derive the performance that can be obtained in such an environment. Results show that, if peers have sufficient information about neighbors, they can be organized in such a way that slow peers obtain near optimal performance without affecting faster peers. On the other hand, if peers do not know neighbor characteristics, slow peers have a significant impact on global performance, and other, more sophisticated, distribution architectures are required to maintain proper scalability
On Some Fundamental Properties of P2P Push/Pull Protocols
No full textThe peer-to-peer communication paradigm is changing the way the Internet works, and the perspective network and service providers are looking at the telecommunication business. P2P applications and networks are taking foot by the day, and new systems are proposed continuously with ever novel features and better performances. In spite of attention and success, however, there is still a lack of fundamental analysis and understanding of the elementary properties of these systems. In this paper we consider a class of P2P protocols suitable both for content delivery (file-based communications) and for high-bandwidth media streaming like video and TV, and explore its fundamental properties. The class considered is known as mesh-based swarming push-pull systems or interleave protocols. They split the content in pieces and alternate continuously two phases: One where the peer pushes a piece to another peer to percolate information in the system, and the other when it pulls a piece trying to retrieve missing information. We compare synchronous and asynchronous models and explore the impact of protocols parameters, such as the dimension of the active neighborhood, trying to identify the efficiency of these very simple protocols in different scenarios, gaining insight to design the next protocol generation with performance and efficiency in min
On the Use of Fixed Point Approximations to Study Reliable Protocols over Congested Links
No full textContent Delivery in Overlay Networks: a Stochastic Graph Processes Perspective
No full textWe consider the problem of distributing a content of finite size to a group of users connected through an overlay network that is built by a peer-to-peer application. The goal is the fastest possible diffusion of the content until it reaches all the peers. Applications like Bit-Torrent or SplitStream are examples where the problem we study is of great interest. In order to represent the content diffusion process, we model the system as a stochastic graph process and define the constraints the graph evolution is subject to. The evolution of the graph is a semi-Markov process where the sojourn times are the rewards of interest for the computation of the time needed to complete the file distribution. We discuss the general properties of the constrained stochastic graphs and we show preliminary results obtained with an ad-hoc Monte-Carlo techniqu
Fast Stochastic Analysis of P2P File Distribution Architectures
No full textn this paper we investigate which is the most efficient architecture and protocol that can be used for file distribution. The focus of the analysis is to understand not only the parameters that influence the distribution process (constraints on the number of neighbors, bandwidth heterogeneity, etc.), but also the impact of the peer behavior, such as selfishness or neighbor selection strategies. The analysis also compares different tree- and mesh-based distribution architectures. We developed an ad-hoc Monte-Carlo technique that is able to analyze scenarios with millions of peers, a network size that traditional discrete-event simulators are not able to treat. The results give an accurate view of the fundamental protocol parameters and policies that impact on the final performance and allow designers to devise improved protocols
Stochastic Graph Processes for Performance Evaluation of Content Delivery Applications in Overlay Networks
No full textThis paper proposes a new methodology to model the distribution of finite-size content to a group of users connected through an overlay network. Our methodology describes the distribution process as a constrained stochastic graph process (CSGP), where the constraints dictated by the content distribution protocol and the characteristics of the overlay network define the interaction among nodes. A CSGP is a semi-Markov process whose state is described by the graph itself. CSGPs offer a powerful description technique that can be exploited by Monte Carlo integration methods to compute in a very efficient way not only the mean but also the full distribution of metrics such as the file download times or the number of hops from the source to the receiving nodes. We model several distribution architectures based on trees and meshes as CSGPs and solve them numerically. We are able to study scenarios with a very large number of nodes, and we can precisely quantify the performance differences between the tree-based and mesh-based distribution architectures
Proximity Detection in Distributed Simulation of Wireless Mobile Systems
No full textThe distributed and the Grid Computing architectures for the simulation of massively populated wireless systems have recently been considered of interest, mainly for cost reasons. Solutions for generalized proximity detection for mobile objects is a relevant problem, with a big impact on the design and the implementation of parallel and distributed simulations of wireless mobile systems. In this paper, a set of solutions based on tailored data structures, new techniques and enhancements of the existing algorithms for generalized proximity detection are proposed and analyzed, to increase the efficiency of distributed simulations. The paper includes the analysis of computation complexity of the proposed solutions and the performance evaluation of a testbed distributed simulation of ad hoc network models. Recent works have shown that the performance of distributed simulation of dynamic complex systems could benefit from a runtime migration mechanism of model entities, which reduces the communication overheads. Such migration mechanisms may interfere with the generalized proximity detection implementations. The analysis performed in this paper illustrates the effects of many possible compositions of the proposed solutions, in a real testbed simulation framework
A Performance Model for Multimedia Services Provisioning on Network Interfaces
Full text linkThis paper presents a method for the performance evaluation of multimedia streaming applications on IP network interfaces with differentiated scheduling. Streaming applications are characterized by the emission of data packets at constant intervals, changing during the connection lifetime, hence a single source can be effectively modeled by an MMDP. We propose an MMDP/D/1/K model to represent the aggregate arrival process and the network interface, assuming that multimedia packet dimensions are approximately constant. A method for solving the above queuing system providing upper and lower bounds to the packet loss rate is presented and results for realistic VoIP applications are discussed and validated against accurate event-driven simulations showing the efficiency and accuracy of the method
Graph based analysis of mesh overlay streaming system
No full textAbstract—This paper studies fundamental properties of stream-based content distribution services. We assume the pres-ence of an overlay network (such as those built by P2P systems) with limited degree of connectivity, and we develop a mathemat-ical model that captures the essential features of overlay-based streaming protocols and systems. The methodology is based on stochastic graph theory, and models the streaming system as a stochastic process, whose characteristics are related to the streaming protocol. The model captures the elementary properties of the streaming system such as the number of active connections, the different play-out delay of nodes, and the probability of not receiving the stream due to node failures/misbehavior. Besides the static properties, the model is able to capture the transient behavior of the distribution graphs, i.e., the evolution of the structure over time, for instance in the initial phase of the distribution process. Contributions of this paper include a detailed definition of the methodology, its comparison with other analytical approaches and with simulative results, and a discussion of the additional insights enabled by this methodology. Results show that mesh based architectures are able to provide bounds on the receiving delay and maintain rate fluctuations due to system dynamics very low. Additionally, given the tight relationship between the stochastic process and the properties of the distribution protocol, this methodology gives basic guidelines for the design of such protocols and systems
Modeling Short-Lived TCP Connections with Open Multiclass Queuing Networks
Full text linkIn this paper we develop an open multiclass queuing network model to describe the behavior of short-lived TCP connections sharing a common IP network. The queuing network model is paired with a simple model of the IP network, and the two models are solved through an iterative procedure. The combined models need as inputs only the primitive network parameters, and they produce estimates of the packet loss probability, the round trip time, the TCP connection throughput, and of the average TCP connection completion time (that is, of the average time necessary to transfer a file with given size over a TCP connection). We derive models for both TCP-Tahoe and TCP-NewReno. The Tahoe model is presented in detail, while the NewReno model is presented describing differences with respect to Tahoe. Results are shown for both models. The analytical performance predictions are validated against detailed simulation experiments in realistic networking scenarios, proving that the proposed modeling approach is accurate
- …
