Towards an Aggregation-Aware Internet Routing

As Internet growth, more and more prefix fragments are announced into the global routing system due to operational reasons of inconsecutive address allocation, multihoming, and traffic engineering. The BGP routing table size in Default Free Zone (DFZ) fast growth will consume more memory space and computational capacity. It has been known that Internet will face with routing scalability issue, especially in the large address space (e.g., IPv6) deployment. In this paper, we propose an innovation to BGP, named Aggregation-aware Inter-Domain Routing (AIDR). It will take the prefix aggregation into account to make tradeoff in the best route selection. We evaluate the effect of AIDR on global routing system using the BGP traces from RouteViews and RIPE. It shows that, averagely, AIDR-based aggregation can reduce to roughly 15%~35% of original routing table size under the 2.0 AS path stretch constraint, and to 25%~40% with no AS path stretch.

Border Gateway Protocol (BGP) is the de facto standard protocol for the inter-domain routing. Due to multi-homing and traffic engineering, the BGP routing table size of default free zone (DFZ) is growing rapidly. Inter-domain routing is facing the scaling challenge. Many solutions have been proposed. Among them, the core-edge separation scheme gets more attentions than others due to its practical advantages. It separates specific prefixes of edge networks from entering into transit core, and reduces the DFZ BGP routing table size. However, there has less evaluation on how much scalability can be improved from core-edge separation. In this paper, we take the further step to quantifying the impact of the core-edge separation on Internet inter-domain routing. We find that separation at stub-transit can reduce 43% routing table size and prevent more than half of BGP updates. We decompose the topology graph by k-core and customer-provider based decomposition methods, and analyze the impact of deploying separation at different level of topological hierarchy. We believe that complicated separation deployment strategies (not the simple stub-transit split) are feasible to approaching an optimal effect.

The Internet routing system faces serious scalability challenges due to the growing number of IP prefixes that needs to be propagated throughout the network. Although IP prefixes are assigned hierarchically and roughly align with geographic regions, today’s Border Gateway Protocol (BGP) and operational practices do not exploit opportunities to aggregate routing information. We present DRAGON, a distributed route-aggregation technique whereby nodes analyze BGP routes across different prefixes to determine which of them can be filtered while respecting the routing policies for forwarding data-packets. DRAGON works with BGP, can be deployed incrementally, and offers incentives for Autonomous Systems (ASs) to upgrade their router software. We illustrate the design of DRAGON through a number of examples, prove its properties while developing a theoretical model of route aggregation, and evaluate its performance. Our experiments with realistic AS-level topologies, assignments of IP prefixes, and routing policies show that DRAGON reduces the number of prefixes in each AS by at least 70% with minimal stretch in the lengths of AS-paths traversed by data packets.

<p>Today, the Internet plays an vital part in our society. We rely greatly on the Internet to work, to communicate and to entertain. The Internet is a very large and complex computer network, consisting of tens of thousands of networks called autonomous systems (ASes). The key routing protocol for interdomain routing between ASes, Border Gateway Protocol (BGP), was invented three decades ago. Although BGP has undergone many improvements, many fundamental problems and limitations of BGP still exist today. For instance, BGP does not have the resiliency to attacks and good support for traffic engineering. To date, many evolutionary and revolutionary solutions have been proposed to address these problems. However, very few were adopted. While there are many reasons for this limited adoption, one may blame for the lack of deployability, scalability and more importantly adequate functionality. SDN is a new networking paradigm that decouples the control plane and data plane. SDN breaks the ossification of the Internet and enables network innovations. SDN has been thoroughly investigated for the enterprise environment. This research investigates the application of SDN for interdomain routing in transit ASes. Specifically, the main goal is to study how SDN capabilities can be utilised to develop a scalable, programmable and flexible routing architecture for transit ISPs.</p>

Traffic engineering is performed by means of a set of techniques that can be used to better control the flow of packets inside an IP network. We discuss the utilization of these techniques across interdomain boundaries in the global Internet. We first analyze the characteristics of interdomain traffic on the basis of measurements from three different Internet service providers and show that a small number of sources are responsible for a large fraction of the traffic. Across interdomain boundaries, traffic engineering relies on a careful tuning of the route advertisements sent via the border gateway protocol. We explain how this tuning can be used to control the flow of incoming and outgoing traffic, and identify its limitations.

Research and discussion on the Internets addressing and routing scalability problems has been for more than a decade. IP aware growing population, and learned needs in terms of site multi-homing, traffic engineering, non-aggregatable address allocations and policy based routing have resulted in the continuous alarming growth of the routing tables in the Default Free Zone (DFZ). Constraints posed by the existing router technology, address allocation and architectural history have made identifying and deploying an effective solution difficult. These sources that influence the core routing table growth can lead to fragmentation or de-aggregation of address prefixes. Much of the growths of the routing tables in the DFZ are a result of de-aggregation of address prefixes. Aggressive topological aggregation is the solution to scale the routing system, that is, IP address allocations should follow topology. IP address allocation and management and the scalability of the routing system are interrelated and only certain IP address allocation and management policies yield scalable routing. The Regional Internet Registries (RIRs) individually use various allocation techniques within their respective regional pools of address space. Proper choice of address allocation algorithm would help in reducing fragmentation, thus reducing routing table size and hence increase Internet addressing and routing scalability. This paper analyzes the various address allocation algorithms used by the RIR’s and based on the lessons learnt proposes a hybrid address allocation scheme for IPv6 that would achieve appreciable address aggregation compared to the existing IPv6 address allocation algorithms, and effectively reduce routing table size in the DFZ. This would help in increasing the Internet addressing and routing scalability and thus enhance the performance of the Internet.

With increasing number of mission critical applications and associated traffic volume on commercial and DoD IP networks, it is becoming increasingly critical for the network infrastructure to be more aware of the quality of service (in a broad sense of the word) requirements and use this awareness in routing, differential treatment of packets, in management-level resource provisioning and allocation. In this paper, we focus on the implication of these requirements on the routing architecture and protocols. In particular, we argue that the inter domain routing protocols like the border gateway protocol (BGP) and its integration with the interior gateway protocol (IGP) need to account for the diversity in the QoS requirements of the applications, users, and mission. The solution we suggest involves allowing multiple inter-AS routes for the same source destination pair, propagation of multiple routes, and of multiple QoS-affecting metrics in the BGP UPDATE messages. These enhancements allow routes to be matched with application requirements and hence allow satisfaction of applications QoS requirements at minimal cost. In order to keep the solution scalable, we limit the propagation of information to non-dominated routes, and design a small set of QoS metrics. Details of the protocols and quantitative analysis will be presented in future papers. Our approach, along with IntServe and DiffServe based QoS capabilities, and traffic engineering extensions of intra domain routing protocols, provide a framework for end-to-end QoS-aware networking.

The Border Gateway Protocol governs the overall routing in the Internet. With the time, it has been overloaded with functions which is was not initially designed for. The main example is Traffic Engineering. Internet Service Providers need to adjust the traffic on their peerings and in absence of a better alternative, use Trial and Error techniques based on manipulating the Autonomous System Path attribute in Border Gateway Protocol (BGP-4) updates. This paper shows sequences of BGP-4 updates where the use of Trial and Error Traffic Engineering techniques have led to instability and the countermeasures used to minimise their impact impact. The analysis methods presented also partially reveal the way Internet Service Providers use AS_PATH Prepending.

The study of Quality of Service (QoS) has become of great importance since the Internet is used to support a wide variety of new services and applications with its legacy structure. Current Internet architecture is based on the Best Effort (BE) model, which attempts to deliver all traffic as soon as possible within the limits of its abilities, but without any guarantee about throughput, delay, packet loss, etc. We develop a three-layer policy based architecture which can be deployed to control network resources intelligently and support QoS sensi-tive applications such as real-time voice and video streams along with standard applications in the Internet. In order to achieve selected QoS parameter values (e.g. loss, delay and PDV) within the bounds set through SLAs for high priority voice traffic in the Internet, we used traffic engineering techniques and policy based routing supported by Border Gateway Protocol (BGP). Use of prototype and simulations validates function-ality of our architecture.

It has been recognized that today's Internet routing and addressing system is facing serious scalability problems. The ever increasing network customers, multihoming, and traffic engineering, are driving the growth of Default Free Zone routing table size at an alarming rate. To solve this issue, several solutions have been proposed to separate the current single numbering space, i.e., the "IP address", into the routing locators' space and the end-systems identifiers' space. In this paper, we first review the recent locator/identifier separation proposals, and give a general classification of them according to their deployment scenarios. We then focus on evaluating the performance of the scalable routing architecture based on the locator/identifier separation. In particular, we explore the improvements that can be achieved in terms of the isolation of providers from customers, the reduction of the global routing table size, and the decrease on the number of routing updates. Furthermore, we evaluate the impact of provider changes on deploying the locator/identifier separation, and suggest that automatic allocation and configuration methods for locator-to-identifier mappings may need to be developed.

Border gateway protocol allows autonomous systems (ASes) to apply diverse routing policies for selecting routes and for propagating reachability information to other ASes. Although a significant number of studies have been focused on the Internet topology, little is known about what routing policies network operators employ to configure their networks. In this paper, we infer and characterize routing policies employed in the Internet. We find that routes learned from customers are preferred over those from peers and providers, and those from peers are typically preferred over those from providers. We present an algorithm for inferring and characterizing export policies. We show that ASes announce their prefixes to a selected subset of providers to perform traffic engineering for incoming traffic. We find that the selective announcement routing policies imply that there are much less available paths in the Internet than shown in the AS connectivity graph, and can make the Internet extremely sensitive to failure events. We hope that our findings will help network operators in designing routing policies.

Multihomed Internet service providers (ISPs) are typically connected to several transit ISPs. If cost is assigned to interconnecting resources, multihomed ISPs face the problem of cost efficient outbound traffic routing. In particular, this is relevant if the ISPs form customer to provider relationships and alternative billing schemes are used, i.e. volume or time based. The border gateway protocol (BGP) is the de facto standard for inter-domain IP routing on the Internet but its traffic engineering capabilities are limited. This paper proposes a local overflow methodology that allows cost efficient outbound packet routing and is based on the scheme for alternative packet overflow routing (SAPOR). The paper identifies the underlying optimisation problem, introduces the modified SAPOR methodology and presents simulation results that verify the operation of the scheme.

The control of multidomain elastic optical networks (EONs) is possible by combining Hierarchical Path Computation Element (H-PCE)-based computation, Border Gateway Protocol with Extensions for Traffic Engineering Link State Information (BGP-LS) topology discovery, remote instantiation via Path Computation Element Communication Protocol (PCEP), and signaling via Resource Reservation Protocol with Extensions for Traffic Engineering (RSVP-TE). Two evolutionary architectures are considered, one based on stateless H-PCE, PCEP instantiation, and end-to-end RSVP-TE signaling (SL-E2E), and a second one based on stateful active H-PCE with perdomain instantiation and stitching. This paper presents the first multiplatform demonstration that fully validates both control architectures achieving multiprotocol interoperability. SL-E2E leads to slightly faster provisioning but needs to keep the state of the stitching of the end-to-end label-switched paths in the parent PCE.

This article considers the growth of the size of the global routing table. The authors attempt to estimate the scale of the growth and its impact for research networks and telecom operators using full-size global routing tables. The article also concerns the impact of the table growth on the stability of the access to the World Wide Web.

Geographic routing is well suited for large scale sensor networks deployments, because the per node state it maintains is independent of the network size. However, due to the “local minimum” caused by holes/obstacles, the path stretch of geographic routing can grow as O(c2), where c is the length of the optimal path. Recently, VIGOR, a geographic routing protocol based on the visibility graph, shows that a constant path stretch can be achieved. This, however, is possible with increased overhead. To address this issue, we propose GOAL (Geometric Routing using Abstracted Holes), a routing protocol that provably achieves a constant path stretch, with lower message, space and computational overhead. We develop a novel distributed convex hull construction (DCC) algorithm that compactly describes holes. This compact representation of a hole is leveraged by nodes to make locally optimal routing decisions. Our theoretical analysis proves the constant stretch property and average stretch of GOAL. Through extensive simulations and a hardware implementation, we demonstrate the effectiveness of GOAL and its feasibility for large-scale sensor networks. In our network settings, GOAL reduces the energy consumption by up to 32%, routing table size by an order of magnitude, when compared with VIGOR.