Time	Title	Speaker	Slides
09:00	Opening	Olav Lysne, Simula
09:10	Fault detection and recovery in the GEANT2 network	Marian Garcia Vidondo, Dante
09:40	Performance Metrics for Resilient Networks	Michael Menth, Wuerzburg University	ppt
10:00	On Practical Deployment of Multi-Topology Based Recovery	George Apostolopoulos, FORTH	ppt
10:20	Coffee break
10:30	IP Fast Reroute with Interface-Specific Forwarding	Srihari Nelakuditi, University of South Carolina	ppt
10:50	IP FRR, what do we have so far?	Amund Kvalbein, Simula	ppt
11:10	IP FRR, what should be done?	Pierre Francois, UCL	pdf
11:30	Inter-domain and LAN Issues in IP FRR	Mike Shand, Cisco	ppt
12:00	ResiliNets: A Franmework for Multilevel Resilient and Survivable Networking	James P.G. Sterbenz, Lancaster University and The University of Kansas	pdf
12:30	Lunch
13:30	Group discussions
	Group 1: Inter-domain issues		ppt
	Group 2: Metrics and framework
	Group 3: Intradomain issues and fast reroute in other network environments (wireless, sensor networks...)		ppt
15:10	Coffee break
15:30	Plenary discussion
17:00	Closing

Talk details

Marian Garcia Vidondo

IP research networks are built using different technologies (IP over SDH, IP over pure Etherent, EoMPLS ...). When all these technologies are stitched together to provide IP transit in a inter-domain environment different aspects need to be considered for a fast fault detection and re-routing. In the presentation we will have a look to the current practices in the GEANT2 network.

Michael Menth

In the presence of local network outages, restoration and protection switching mechanisms redirect the traffic over alternative paths to mitigate the effect of failures. However, some failure combinations do still lead to loss of end-to-end (e2e) connectivity and to severe congestion due to rerouted traffic. Congestion may also be caused by unexpected traffic shifts due to changed user behavior or due to changes of interdomain routing. This work presents a framework for the analysis of e2e unavailability and congestion due to (1) failures, (2) changes of user behavior, and (3) changed interdomain routing. The analysis finds most probable combinations of (1), (2), and (3), and evaluates the connectivity and the link utilization of the network under these conditions. We have implemented this concept in a software tool and its visualization of the results leads to a comprehensive view of the network~Rs resilience. It helps to anticipate potential e2e disconnection and congestion before failures and overload occur or before planned modifications (new infrastructure, new routing, new customers) take effect. Thus, it detects weak points in a network, predicts the effectiveness of potential upgrades, and supports thereby careful bandwidth overprovisioning.

George Apostolopoulos

We discuss issues that have to do with the practical deployment of a multi-topology based recovery mechanisms for IP networks (multiple areas, inter-AS link failures) and related issues as well as extending this approach to MPLS networks.

Srihari Nelakuditi

Traditional IP networks forward a packet based on destination address alone. Instead, we propose to determine the next-hop of a packet based on both its destination and also previous-hop/incoming-interface. In this talk, we argue that such an interface-specific forwarding enables loop-free fast-rerouting with minimal changes to the forwarding plane of the Internet.

Amund Kvalbein

We briefly discuss the properties of Not-via, FIR and MRC with respect to some selected metrics, and point out what we believe to be the pros and cons for each of the methods. We hope this can spur some discussion!

Pierre Francois

The presentation will try to bring questions on what should (not) be done for IP-FRR.

Mike Shand

• Extension of the not-via technique to other than Link State protocols (such as Distance Vector, BGP etc.).
• Loop free convergence techniques for BGP
• Dealing with multiple routing domains, including interactions between them. (e.g. multiple link state areas, or multiple ASs and the combination of link state and BGP domains.
• Dealing with LANs. LANs do not scale particularly well in most IPFRR solutions, particularly when doing node protection.

James P G Sterbenz

We have developed a framework for resilient networking that aims to unify the fields of fault-tolerance, dependability, and survivability and is based upon a strategy D^2R^2+DR: defend, detect, remediate, recover + diagnose, refine. We have developed a set of architectural principles for resilient network design to realise this strategy. We will briefly present the overall framework, and then discuss how IP rerouting fits. In particular, IP rerouting is only part of the puzzle, situated at one of the levels (network layer) and part of the strategy (remediation and recovery).