Elsevier

Computer Networks

Volume 49, Issue 4, 15 November 2005, Pages 593-619
Computer Networks

Inter-domain routing: Algorithms for QoS guarantees

https://doi.org/10.1016/j.comnet.2005.01.017Get rights and content

Abstract

Quality-of-Service routing satisfies performance requirements of applications and maximizes utilization of network resources by selecting paths based on the resource needs of application sessions and link load. QoS routing can significantly increase the number of reserved bandwidth sessions that a network can carry, while meeting application QoS requirements. Most research on QoS routing to date, has focused on routing within a single domain. BGP, the de facto standard for inter-domain routing provides no support for QoS routing, and has well-documented performance related issues that lead to its inadequacy to support QoS. This paper proposes new approaches to inter-domain routing for sessions requiring guaranteed QoS. The performance-scalability tradeoff is explored via extensive experiments on the proposed algorithms. Our extensive experiments on realistic intra-domain ISP topologies as well as inter-domain settings, show that the proposed algorithms achieve at least an order of magnitude gain in performance (blocking probability) over current mechanisms, while remaining scalable and easy to deploy.

Introduction

The need for timely delivery of real-time information over local and wide area networks is becoming more common due to the rapid expansion of the Internet user population in recent years, and the growing interest in using the Internet for telephony, video conferencing and other multimedia applications. Choosing a route that meets the resource needs of such applications is essential to the provision of the high quality services that users are coming to expect.

In this context, it is important to distinguish datagram and flow routing. In datagram routing, packets of a session may follow different paths to the destination. In flow routing, all packets belonging to an application session follow the same path, allowing bandwidth to be reserved along that path, in order to ensure high quality of service. Because many thousands or even millions of packets are typically sent during a single application session, flow routing occurs far less often than datagram routing, making it practical to apply more complex decision procedures than can be used in datagram routing.

The current Internet follows the datagram routing model and relies on adaptive congestion control to cope with overloads. Internet traffic is forwarded on a best-effort basis with no guarantees of performance. This can result in wide variations in performance, resulting in poor service quality for applications such as voice and video. Furthermore, Internet routing is typically topology-driven instead of being load-driven. This approach does not allow traffic to be routed along alternative paths, when the primary route to a destination becomes overloaded. While the application of load-sensitive routing to datagram traffic can cause hard-to-control traffic fluctuations, it can be successfully applied to flow routing, since reserved bandwidths sessions typically have holding times of minutes, effectively damping any rapid fluctuations in routes. Note that the use of the term “flow” in the rest of this paper also applies to aggregates of smaller “micro-flows” which are often bundled, when routing across domains.

The most prominent inter-domain routing protocol in the current Internet is the Border Gateway Protocol (BGP) [1]. BGP is a path vector based protocol, where a path refers to a sequence of intermediate domains between source and destination routers. BGP suffers from a number of well-documented problems, including long convergence times [2], [3] following link failures. BGP adopts a policy based routing mechanism whereby each domain applies local policies to select the best route and to decide whether or not to propagate this route to neighboring domains without divulging their policies and topology to others.

The immediate effect of the policy based approach is to potentially limit the possible paths between each pair of Internet hosts. BGP does not ensure that every pair of hosts can communicate even though there may exist a valid path between the hosts. Also, since every domain is allowed to use its own policy to determine routes, the final outcome may be a path that is locally optimal at some domains but globally sub-optimal due to the lack of a uniform policy or metric used to find an end-to-end route. This point is highlighted by [4], [5], where a majority of paths that are picked by BGP do not represent the optimal end-to-end paths. The authors define “optimal paths” by hop count in [4]. Their results show that for 50% of the BGP paths, there exists an alternate path with at least 5 less hops. In [5], different measures of path quality such as loss rate, bandwidth and round-trip time, consistently indicate that 30-80% of paths actually have an alternate path with significantly superior quality than the default path chosen by BGP.

The sub-optimality of BGP is primarily due to most domains defaulting to hot potato routing, in which each domain in the end-to-end path, tries to shunt packets as quickly as possible to the next network in the path, rather than selecting routes that will produce the best end-to-end performance for users. This characteristic is clearly undesirable, even for datagram traffic, and is particularly problematic for sessions that require high quality of service. Thus, there is clearly a critical need for a QoS routing mechanism that allows guarantees across domains.

Most research in QoS routing has focused on routing within a single domain. While the intra-domain problem is important, it is arguably even more important to address the QoS routing problem at the inter-domain level. Also, it is not feasible to directly extend protocols for intra-domain routing to the inter-domain context. While scalability is an even larger concern due to the sheer number of nodes and domains, the issue of peering relationships can constrain the nature and periodicity of information exchanged between domains. Whereas, an intra-domain protocol operates in a smaller network where all routers cooperate such that information about the entire topology can be conveyed to each router, this is not possible when routing across domains. Providing QoS routes across domains is made harder by the fact that each domain has a constraint on the information that it exchanges with other domains. This can affect the routes that are advertized by BGP speakers and can also isolate domains even though valid paths exist between them. Thus, there is a necessity to develop QoS routing approaches that are backward compatible with current networks while using simple mechanisms to avoid scalability problems and to facilitate deployment.

New inter-domain QoS routing approach: We begin with the observation that the peering links that connect distinct routing domains are typically the primary congestion points or bottlenecks for network traffic [6].1 Managing the resource use at such points of congestion is clearly critical to providing end-to-end quality of service.

Our strategy for inter-domain routing has two parts. In the inter-domain part, a loose source route2 is selected by the router at origination point of the session. This source route specifies the domains through which the route is to pass and the peering links used to pass from one domain to the next. Within each domain, paths are selected between the ingress and egress points, using domain-specific routing policies. This is referred to as the intra-domain part. This decomposition of the end-to-end routing problem respects each domain’s right to maintain the privacy of its internal network configuration and appropriately limits the amount of information that must be taken into account when selecting routes. At the same time, it allows the large-scale characteristics of the route to be selected with appropriate consideration of the status of the peering links.

Privacy issues: While it may seem a violation of privacy to disseminate the peering link information, we believe that it is in the provider’s best interest with several beneficial consequences. For example, providing information about a chronically saturated peering link may mitigate the load on that link. Balancing the load not only increases the QoS to the client, but also creates more revenue to the participating providers (on the end-to-end path), since they are now able to carry more calls with less rejection probability. We discuss this in more detail in Section 6.

Scalability issues: With any QoS routing scheme, scalability is an important criteria. In terms of a real operating environment with thousands of autonomous systems, there will obviously be severe scaling issues if all the peering link information needs to be disseminated. We discuss this issue and its potential solution in more detail in Section 6.

Contributions: In this paper, we propose and evaluate two inter-domain QoS routing algorithms that follow the strategy of improving performance by avoiding bottle-neck peering links. We begin by describing a simple overlay network architecture to enable the QoS routing mechanisms such that no change is necessary to existing mechanisms such as BGP. The first, uses a fairly conventional shortest-path framework, using a cost metric that accounts for both the intrinsic cost of each link and the amount of bandwidth that the link has available for use. The second dynamically probes several paths in parallel, in order to find a path capable of handling the flow. This approach eliminates the need for regular routing updates, since routing information is obtained on-demand. We perform extensive experiments on realistic intra-domain and inter-domain topologies to demonstrate the substantial benefits that can be attained by using our proposed schemes.

There is a significant amount of prior research in the area of intra-domain QoS routing. However, the design criteria for inter-domain QoS routing is different from intra-domain routing, with a special emphasis on scalability. There is an inherent tradeoff between performance and scalability. We believe that this is one of the first papers that extensively evaluates the performance of QoS routing protocols in both the intra and inter-domain context and quantitatively evaluates the tradeoff.

The outline of the paper is as follows. Section 2 motivates and describes two QoS routing algorithms that are extended for inter-domain routing. Performance of these routing algorithms in an intra-domain setting are also presented to demonstrate their benefits. Section 3 extends these algorithms to support inter-domain QoS. Section 4 describes the inter-domain simulation environment and a comprehensive evaluation of the performance of the routing algorithms including a comparison with a BGP based approach. Section 5 enhances the QoS routing algorithms in order to facilitate deployment in current networks. Implementations details along with the processing overheads of deploying the proposed algorithms are presented in Section 6. Section 7 presents prior research in the area and the paper concludes in Section 8.

Section snippets

Motivating new QoS routing algorithms

To ensure the best possible performance in the presence of the full range of traffic conditions, it is important for route selection to be guided by a knowledge of current network state information using metrics such as loss, latency and available link capacity. Available bandwidth in particular is especially critical at the inter-domain level where peering link bottlenecks are critical in determining whether a packet reaches the destination. In this paper, we focus on using available bandwidth

Inter-domain QoS routing algorithms

The objective of the proposed inter-domain routing algorithm is to select a loose source route, joining a flow’s endpoints. Since peering links are often congestion points for network traffic, careful selection of peering links can have a significant impact on the probability of success. Because the inter-domain route selection must be done without the benefit of detailed knowledge of each domain’s internal configuration, it’s necessary to estimate the amount resources or the cost incurred by a

Results for inter-domain QoS routing

In this section, we will first describe the design of an inter-domain topology to evaluate the performance of the inter-domain QoS routing algorithms. We then describe extensive results on the performance of the LCC and PP versions along with enhancements to facilitate their deployment.

Enhancing the routing algorithms

We will now describe enhancements to the aforementioned QoS routing algorithms that explore the tradeoff between performance and algorithm complexity/scalability. We have the PP and GEO algorithms at two extremes. The PP algorithm shows the best performance, but also incurs a size-able overhead due to probes being sent, as well as requiring that all domains use the algorithm consistently for inter and intra domain routing on the end-to-end path. Both of the above do not make the algorithm

Implementation of routing algorithms

There are several assumptions made by the routing algorithms presented in this paper that would not normally be realized in current networks. Some of the mechanisms that would be barriers are discussed below.

Related work

In addition to the work mentioned in Section 2, there has been substantial amount of prior research on QoS routing within a domain. We summarize a few of the representative schemes below and examine the complexity and scalability issues that typically act as a barrier to deployment. In particular, we focus on multi-path routing schemes since the PP algorithm proposed in this paper is the best performing algorithm.

There has also been substantial prior research on multi-path routing. One such

Conclusions

In this paper, we studied the performance of various QoS routing schemes that cover a wide spectrum from shortest path routing to hybrid multi-path routing schemes. These algorithms are enabled using an overlay network architecture which enables peering link state to be disseminated without relying on changes to existing protocols such as BGP. The architecture also allows flexible injection of metrics that could be used for QoS routing including available bandwidth, and delay, depending on the

Samphel Norden received a B.S. (1998) from Indian Institute of Technology, Madras and Doctor of Science (D.Sc.) (2002) degrees in Computer Science from Washington University in St. Louis. He is currently a Member of Technical Staff (MTS) in the Center for Mobile Networking Research in Lucent Bell Laboratories. His research interests include Mobile Networking, Denial-of-Service detection and prevention, Inter-domain QoS routing, Overlay Networks and Wireless Security.

References (31)

  • J.A. Fingerhut et al.

    Designing least-cost nonblocking broadband networks

    J. Algorithms

    (1997)
  • Y. Rekhter, T. Li, A Border Gateway Protocol 4 (BGP-4), Internet RFC-1771, March...
  • C. Labovitz, A. Ahuja, A. Bose, F. Jahanian, Delayed internet routing convergence, in: Proceedings of ACM SIGCOMM,...
  • J. Garcia-Luna-Aceves

    Loop-free routing using diffusing computations

    IEEE/ACM Trans. Networking

    (1993)
  • H. Tangmunarunkit, R. Govindan, S. Shenker, D. Estrin, The impact of routing policy on internet paths, in: Proceedings...
  • S. Savage, A. Collins, E. Hoffman, J. Snell, T. Anderson, The end-to-end effects of internet path selection, In...
  • Akamai Inc. Internet Bottlenecks. Akamai Inc. White paper,...
  • Q. Ma, P. Steenkiste, Quality-of-service routing for traffic with performance guarantees, in: Proceedings of IFIP...
  • I. Matta, A.U. Shankar, Dynamic routing of real-time virtual circuits. in: Proceedings IEEE International Conference on...
  • A. Shaikh et al.

    Evaluating the impact of stale link state on quality-of-service routing

    IEEE/ACM Trans. Networking

    (2001)
  • D. Katz, D. Yeung, Traffic engineering extensions to OSPF, Internet Engineering Task Force, July 1997. Internet...
  • S. Norden, Improving network performance using QoS routing and deferred reservations. Ph.D. thesis, Department of...
  • R. Guerin, A. Orda, D. Williams, QoS routing mechanisms and OSPF extensions, in: Proceedings of IEEE INFOCOM, March...
  • G. Apostolopoulos, R. Guerin, S. Kamat, A. Orda, T. Przygienda, D. Williams, QoS Routing mechanismsn and OSPF...
  • X. Liu, K. Ravindran, Single-hop probing asymptotics in available bandwidth estimation: sample path analysis, in:...
  • Cited by (0)

    Samphel Norden received a B.S. (1998) from Indian Institute of Technology, Madras and Doctor of Science (D.Sc.) (2002) degrees in Computer Science from Washington University in St. Louis. He is currently a Member of Technical Staff (MTS) in the Center for Mobile Networking Research in Lucent Bell Laboratories. His research interests include Mobile Networking, Denial-of-Service detection and prevention, Inter-domain QoS routing, Overlay Networks and Wireless Security.

    View full text