Francois Clad

Graceful Convergence in Link-State IP Networks: A Lightweight Algorithm Ensuring Minimal Operational Impact

The use of real-time multimedia or mission-critical applications over IP networks puts strong pressure on service providers to operate disruption-free networks. However, after any topological change, link-state Interior Gateway Protocols (IGPs), such as IS-IS or OSPF, enter a convergence phase during which transient forwarding loops may occur. Such loops increase the network latency and cause packet losses. In this paper, we propose and evaluate an efficient algorithm aimed at avoiding such traffic disruptions without modifying these IGPs. In case of an intentional modification of the weight of a link (e.g., to shut it down for maintenance operations or to perform traffic engineering), our algorithm iteratively changes this weight, splitting the modification into a sequence of loop-free transitions. The number of weight increments that need to be applied on the link to reach its target state is minimized in order to remain usable in existing networks. Analysis performed on inferred and real Internet service provider (ISP) topologies shows that few weight increments are required to handle most link shutdown events (less than two intermediate metrics for more than 85% of the links). The evaluation of our implementation also reveals that these minimal sequences can be computed in a reasonable time.

Graceful router updates in link-state protocols

Manageability and evolvability are crucial needs for IP networks. Unfortunately, planned topological changes may lead to transient forwarding loops in link-state routing protocols commonly used in IP networks. These lead to service unavailability, reducing the frequency at which operators can adapt the network topology. Prior works proved that the state of a given link can be modified while avoiding forwarding inconsistencies without changing protocol specifications. In this paper, we study the more general problem of gracefully modifying the state of an entire router, while minimizing the induced operational impact. As opposed to a single-link modification, the router update problem is k-dimensional for a node of degree k. Moreover, we show that the interplay between operations applied at the router granularity can lead to loops that do not occur considering a single-link modification. In this paper, we present an efficient algorithm that computes minimal sequences of weights to be configured on the links of the updated node. Based on real IP network topologies, we show that the size of such sequence is limited in practice.

Computing minimal update sequences for graceful router-wide reconfigurations

Manageability and high availability are critical properties for IP networks. Unfortunately, with link-state routing protocols commonly used in such networks, topological changes lead to transient forwarding loops inducing service disruption. This reduces the frequency at which operators can adapt their network. Prior works proved that it is possible to avoid disruptions due to the planned reconfiguration of a link by progressively changing its weight, leading to a solution that does not require changing protocol specification. In this paper, we study the more general problem of gracefully modifying the logical state of multiple interfaces of a router, while minimizing the number of weight updates. Compared to single-link modifications, the router update problem is k-dimensional for a router having k neighbors. We also show that multidimensional updates may trigger new kinds of disruptions that make the problem more challenging than the single-link case. We then present and evaluate efficient algorithms that compute minimal sequences of weights enabling disruption-free router reconfigurations. Based on analysis of real IP network topologies, we show that both the size of such sequences and the computing time taken by our algorithms are limited.

Energy-efficient data collection in WSN: A sink-oriented dynamic backbone

In wireless sensor networks, energy efficiency is generally achieved by turning off some capabilities from a subset of deployed sensors. The set of active nodes must therefore meet the application requirements (e.g. area coverage, data redundancy) while remaining fully connected to allow further data collection. Here, we focus on the case of a nomad sink entering the network and gathering every monitoring data. At the routing layer, minimizing the number of nodes acting as relays requires to construct a maximum leaf spanning tree (MLST). However, optimizing convergecast communications consists in minimizing the hop distance between the sink and all others nodes, leading so to a shortest path tree rooted at the sink. In this paper, we propose a distributed routing protocol that aims at constructing an energy efficient backbone being convergecast efficient at the same time. Our proposal introduces a tradeoff parameter to adjust the compromise “number of relays / routing efficiency” and then constructs a hybrid routing structure based on the combination of variants of the Wu-Li algorithm and a gradient-based routing protocol. For all topologies we simulated, and when tuned for energy saving, our approach outperforms a 2-approximation for constructing a MLST. Furthermore, when tuned for convergecast routing, simulation results show that our solution constructs a routing optimal backbone that involves a small fraction of relays.

Software Resolved Networks: Rethinking Enterprise Networks with IPv6 Segment Routing

Enterprise networks often need to implement complex policies that match business objectives. They will embrace IPv6 like ISP networks in the coming years. Among the benefits of IPv6, the recently proposed IPv6 Segment Routing (SRv6) architecture supports richer policies in a clean manner. This matches very well the requirements of enterprise networks. In this paper, we propose Software Resolved Networks (SRNs), a new architecture for IPv6 enterprise networks. We apply the fundamental principles of Software Defined Networks, i.e., the ability to control the operation of the network through software, but in a different manner that also involves the endhosts. We leverage SRv6 to enforce and control network paths according to the network policies. Those paths are computed by a centralized controller that interacts with the endhosts through the DNS protocol. We implement a Software Resolved Network on Linux endhosts, routers and controllers. Through benchmarks and simulations, we analyze the performance of those SRNs, and demonstrate that they meet the expectations of enterprise networks.

Implementation of virtual network function chaining through segment routing in a linux-based NFV infrastructure

This paper presents an architecture to support Vir- tual Network Functions (VNFs) chaining using the IPv6 Segment Routing (SR) network programming model. Two classes of VNFs are considered: SR-aware and SR-unaware. The operations to support both SR-aware and SR-unaware VNFs are described at an architectural level and we propose a solution for SR-unaware VNFs hosted in a NFV node. An Open Source implementation of the proposed solution for a Linux based NFV host is available and a set of performance measurements have been carried out in a testbed.

A fine-grained multi-source measurement platform correlating routing transitions with packet losses

Abstract In this paper, we are interested in the relationship between packet losses and routing changes in an operational network. To do so we designed and deployed DCART, a monitoring platform over RENATER, the French research and education network. Our platform collects four data sources using both active and passive measurements in order to unveil their temporal correlations. Active probing allows especially for measuring packet losses on specifically crafted data flows. Those flows explore several load balanced paths and ease the revelation of forwarding loops. Passive monitoring is achieved by listening to all routing updates from IS-IS, the intra-domain routing protocol in use, and by retrieving tickets generated by the Network Operations Center (NOC). During our monitoring campaign, we observe that most of the series of loss were correlated to routing events either because routing changes lead to inconsistent state transitions, or because faulty – and so lossy – links trigger numerous periods of link flapping. In particular, we show that losses due to forwarding loops resulting from inconsistent routing states are quite common when links come back after an outage. We also show that link flapping sometimes induce very long lasting lossy periods frequently unnoticed by the NOC. A lightweight monitoring platform such as DCART could be used to better anticipate recurrent network outages and to improve the ticketing system.