Assessment, design and development of novel inter- and intradomain topology discovery tools

The capability of processing routing information and generating network topology maps which can be used as inputs for network models is extremely useful when planning fail-over and network redundancy strategies.

In the interdomain or inter-provider field, many projects [66] have shown the possibility of using information gathered by public BGP-4 route repositories [67] for building high level topology maps. Some tools like BGPlay [68] use this information to show the evolution in time of the connectivity between providers.

These high level topology maps show the providers and the logical links between them with limited information about the actual physical links or routing devices. Nevertheless, European Research Projects, like INTERMON [69], have shown, that this information is enough to establish causal links between anomalies in Quality of Service patterns and interprovider routing events like

-      Denial of Service Attacks on the border routers,

-      unintentional configuration errors or configuration by trial and error methodologies.

Next steps in interdomain routing information processing include the automated detection of network configurations which are at the origin of 'BGP Wedgies', or bad routing policy interactions that cannot be debugged [32]. These interactions could lead to situations where a network segment consisting of several network providers doesn't recover as expected from a network failure.

Current technologies to discover interdomain paths, which include information of all routers traversed between an origin and a destination are based on active network probes and ICMP responses, which are derivate work from traceroute [33]. The active network probing technologies have a common drawback: they introduce additional traffic in the network and they have already been misused to produce Distributed Denial of Service Attacks. Additionally, with the introduction of the MPLS technology in the network backbones, the information gathered by the probes is not always accurate.

INTERSECTION proposes an evolutionary path to extend the route collector paradigm to include MPLS networks which covers design and proof of concept for the following environments:

–interior IP routing and forwarding using MPLS

–advanced MPLS services (Layer 2 and 3 VPNs)

–native MPLS interprovider connections

–route collectors for Generalised MPLS (GMPLS) networks

Route collectors have had some success because they build on a common specification for the storage of routing event information. It is based on the binary format specified by Merritt's MRT (multithreaded routing toolkit), which was taken over by the IETF GROW working group and is now an Internet Draft [34]. INTERSECTION will participate in the definition process of the successors of the current MRT binary format specification.

INTERSECTION also proposes to employ the new visualisation framework (see section 1.2.6) to display the collected routing information. To enable this, an interface between the route collector and visualisation tools will be designed and implemented.

References

[32] T. Griffin, "BGP Wedgies: Bad Routing Policy Interactions that Cannot be Debugged", Proceedings of IDRWS, Amsterdam, May 2004

[33] G. Malkin, ”Traceroute Using an IP Option”, IETF RFC 1393

[34]L. Blunk,M. Karir,Merit Network,C. Labovitz,”MRT routing information export format”, draft-ietf-grow-mrt-04.txt

[66]The Oregon Routeviews Project: http://www.routeviews.org 

[67] The RIPE-RIS database http://www.ripe.net/projects/ris/index.html

[68] BGPlay, http://www.ris.ripe.net/bgplay

[69] IST-INTERMON: Advanced architecture for INTER-domain quality of service MONitoring, modelling and visualisation (IST-2001-34123) http://www.ist-intermon.org

Back to Project Detail