Use this URL to cite or link to this record in EThOS:
Title: Provisioning IP backbone networks based on measurements
Author: Papagiannaki, Konstantina
ISNI:       0000 0001 3466 9683
Awarding Body: University of London
Current Institution: University College London (University of London)
Date of Award: 2003
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The theme of this thesis is the enhancement of current IP backbone provisioning practices in the presence of additional network measurements. Current practices are heavily dependent on the intuition of the human operators. Traffic variability, scalability issues, lack of monitoring information, and complex interactions between inter- and intra-domain routing protocols result in network management techniques that usually rely on trial and error. In contrast with reductionist approaches, we demonstrate the benefits of using different types of monitoring information in the formalisation of different network provisioning tasks, and provide a methodological framework for their analysis. We use four main sources of network monitoring information: (i) GPS-synchronised packet traces listing every packet traversing a monitored unidirectional link, (ii) BGP routing table dumps, (iii) SNMP information collected since 1999, and (iv) topological information. Combining the above sources of information, and analysing them at the appropriate time scale, we demonstrate the benefits of additional measurements on three specific network provisioning tasks. First, we measure and analyse delay as experienced by packets while traversing a single router inside the network. We show that packets experience minimal queueing delay and that delay through the network is dominated by the propagation delay. Our results hold when network link utilisation stays moderate. However, links are likely to experience short-lived congestion episodes as a result of link or equipment failures. Our second network provisioning task regards the off-loading of congested links by the re-direction of high-volume flows. We propose a methodology for the identification of those flows traversing a link that contribute significant amounts of traffic consistently over time. Persistent link overload can only be resolved through additional provisioning. Our third task focuses on the prediction of where and when future provisioning will be required in the backbone. We obtain accurate predictions for at least six months in the future.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available