This paper introduces AntNet, a novel approach to the adaptive learning of routing tables in communications networks. AntNet is a distributed, mobile agents based Monte Carlo system that was inspired by recent work on the ant colony metaphor for solving optimization problems. AntNet's agents concurrently explore the network and exchange collected information. The communication among the agents is indirect and asynchronous, mediated by the network itself. This form of communication is typical of social insects and is called stigmergy. We compare our algorithm with six state-of-the-art routing algorithms coming from the telecommunications and machine learning elds. The algorithms' performance is evaluated over a set of realistic testbeds. We run many experiments over real and arti cial IP datagram networks with increasing number of nodes and under several paradigmatic spatial and temporal tra c distributions. Results are very encouraging. AntNet showed superior performance under all the experimental conditions with respect to its competitors. We analyze the main characteristics of the algorithm and try to explain the reasons for its superiority. 1.

Managing large IP networks requires an understanding of the current traffic ows, routing policies, and network configuration. Yet, the state-of-the-art for managing IP networks involves manual con guration of each IP router, and traffic engineering based on limited measurements. The networking industry is sorely lacking in software systems that a large Internet Service Provider (ISP) can use to support traffic measurement and network modeling, the underpinnings of effective traffic engineering. This paper describes the AT&T Labs NetScope, a unified set of software tools for managing the performance of IP backbone networks. The key idea behind NetScope is to generate global views of the network, on the basis of configuration and usage data associated with the individual network elements. Having created an appropriate global view, we are able to infer and visualize the network-wide implications of local changes in traffic, con guration, and control. Using NetScope, a network provider can experiment with changes in network configuration in a simulated environment, rather than the operational network. In addition, the tool provides a sound framework for additional modules for network optimization and performance debugging. We demonstrate the capabilities of the tool through an example traffic-engineering exercise of locating a heavily-loaded link, identifying which traffic demands flow on the link, and changing the configuration of intra-domain routing to reduce the congestion.

An IP routing protocol is safe if it is guaranteed to converge in the absence of network topology changes. BGP, currently the only interdomain routing protocol employed on the Internet, is not safe in this sense. It may seem that the source of BGP's potential divergence is inherent in the requirements for any interdomain routing protocol --- policy-based metrics must be allowed to override distance-based metrics, and each autonomous system must be allowed to independently define its routing policies with little or no global coordination. In this paper we present a Simple Path Vector Protocol (SPVP) that captures the underlying semantics of BGP by abstracting away all nonessential details. We then add a dynamically computed attribute to SPVP routing messages, called the route history. Protocol oscillations caused by policy conflicts produce routes whose histories contain cycles. These cycles identify the policy conflicts and the autonomous systems involved. SPVP is made safe by automati...

The smooth operation of the Internet depends on the careful configuration of routers in thousands of autonomous systems throughout the world. Configuring routers is extremely complicated because of the diversity of network equipment, the large number of configuration options, and the interaction of configuration parameters across multiple routers. Network operators have limited tools to aid in configuring large backbone networks. Manual configuration of individual routers can introduce errors and inconsistencies with unforeseen consequences for the operational network. In this paper, we describe how to identify configuration mistakes by parsing and analyzing configuration data extracted from the various routers. We first present an overview of IP networking from the viewpoint of an Internet Service Provider (ISP) and describe the kinds of errors that can appear within and across router configuration files. To narrow the scope of the problem, we then focus our attention on the configuration commands that relate to traffic engineering---tuning the intradomain routing protocol to control the flow of traffic through the ISP network. We present a case study of a prototype tool, developed in collaboration with AT&T IP Services, for checking the configuration of the AT&T IP Backbone and providing input to other systems for network visualization and traffic engineering. 1

Prompted by the advent of QoS routing in the Internet, we investigate the properties that path weight functions must have so that hop-by-hop routing is possible and optimal paths can be computed with a generalized Dijsktra's algorithm. For this purpose we define an algebra of weights which contains a binary operation, for the composition of link weights into path weights, and an order relation. Isotonicity is the key property of the algebra. It states that the order relation between the weights of any two paths is preserved if both of them are either prefixed or appended by a common, third, path. We show that isotonicity is both necessary and sufficient for a generalized Dijkstra's algorithm to yield optimal paths. Likewise, isotonicity is also both necessary and sufficient for hop-by-hop routing. However, without strict isotonicity, hop-by-hop routing based on optimal paths may produce routing loops. They are prevented if every node computes what we call lexicographic-optimal paths. These paths can be computed with an enhanced Dijkstra's algorithm that has the same complexity as the standard one. Our findings are extended to multipath routing as well. As special cases of the general approach, we conclude that shortestwidest paths can neither be computed with a generalized Dijkstra's algorithm nor can packets be routed hop-by-hop over those paths. In addition, loop-free hop-by-hop routing over widest and widest-shortest paths requires that each node computes lexicographic-optimal paths, in general.

This paper presents BCube, a new network architecture specifically designed for shipping-container based, modular data centers. At the core of the BCube architecture is its server-centric network structure, where servers with multiple network ports connect to multiple layers of COTS (commodity off-the-shelf) mini-switches. Servers act as not only end hosts, but also relay nodes for each other. BCube supports various bandwidth-intensive applications by speedingup one-to-one, one-to-several, and one-to-all traffic patterns, and by providing high network capacity for all-to-all traffic. BCube exhibits graceful performance degradation as the server and/or switch failure rate increases. This property is of special importance for shipping-container data centers, since once the container is sealed and operational, it becomes very difficult to repair or replace its components. Our implementation experiences show that BCube can be seamlessly integrated with the TCP/IP protocol stack and BCube packet forwarding can be efficiently implemented in both hardware and software. Experiments in our testbed demonstrate that BCube is fault tolerant and load balancing and it significantly accelerates representative bandwidthintensive applications.

IP networks today require massive effort to configure and manage. Ethernet is vastly simpler to manage, but does not scale beyond small local area networks. This paper describes an alternative network architecture called SEATTLE that achieves the best of both worlds: The scalability of IP combined with the simplicity of Ethernet. SEATTLE provides plug-and-play functionality via flat addressing, while ensuring scalability and efficiency through shortest-path routing and hash-based resolution of host information. In contrast to previous work on identity-based routing, SEAT-TLE ensures path predictability and stability, and simplifies network management. We performed a simulation study driven by real-world traffic traces and network topologies, and used Emulab to evaluate a prototype of our design based on the Click and XORP open-source routing platforms. Our experiments show that SEAT-TLE efficiently handles network failures and host mobility, while reducing control overhead and state requirements by roughly two orders of magnitude compared with Ethernet bridging.

Abstract not found

The OSPF and IS-IS routing protocols widely used in today's Internet compute a shortest path tree (SPT) from each router to other routers in a routing area. Many existing commercial routers recompute an SPT from scratch following changes in the link states of the network. Such recomputation of an entire SPT is inecient and may consume a considerable amount of CPU time. Moreover, as there may coexist multiple SPTs in a network with a set of given link states, recomputation from scratch causes frequent unnecessary changes in the topology of an existing SPT and may lead to routing instability. In this paper, we present new dynamic SPT algorithms that make use of the structure of the previously computed SPT. Besides efficiency, our algorithm design objective is to achieve routing stability by making minimum changes to the topology of an existing SPT (while maintaining shortest path property) when some link states in the network have changed. We establish an algorithmic framework that allows ...

Distributed path restoration based on optical cross-connects can provide highly capacity-efficient real time restoration for WDM-based optical networking. However, to obtain an assured restoration level with the theoretically very low amounts of spare capacity that path restoration allows one must solve, or closely approximate a solution to, the integer multicommodity maximum flow (MCMF) problem. MCMF is, however, a hard combinatorial optimization problem due to what is called the "mutual capacity" aspects of the problem: which of many competing Origin-Destination pairs should be allowed paths over the finite spares on each span? Integer MCMF is further complicated by the non uni-modular nature of the problem, i.e. fractional flows are forbidden but would arise if solved by Linear Programming. This paper presents a heuristic principle that tests well against Integer Programming solutions of MCMF routing. The heuristic is first characterized in a centralized program, then adapted for use in a distributed path restoration protocol. In all test cases the protocol obtains over 97% of the paths found in an optimal MCMF solution in the same network. Via OPNET simulation it is also predicted that the protocol will run in well under 2 seconds which means it could be used directly in real-time, or in distributed pre-failure self-planning, for restoration. The significance is that network operators could aggressively optimize their spare capacity, towards theoretical minimums, while still assuring 100% restorability.