Engineering a large IP backbone network without an accurate network-wide vi
ew of the traffic demands is challenging. Shifts in user behavior, changes
in routing policies, and failures of network elements can result in signifi
cant (and sudden) fluctuations in load. In this paper, we present a model o
f traffic demands to support traffic engineering and performance debugging
of large Internet Service Provider networks. By defining a traffic demand a
s a volume of load originating from an ingress link and destined to a set o
f egress links, we can capture and predict how routing affects the traffic
traveling between domains. To infer the traffic demands, we propose a measu
rement methodology that combines Bow-level measurements collected at all in
gress links with reachability information about all egress links. We discus
s how to cope with situations where practical considerations limit the amou
nt and quality of the necessary data, Specifically, we show how to infer in
terdomain traffic demands using measurements collected at a smaller number
of edge links-the peering links connecting to neighboring providers. We rep
ort on our experiences in deriving the traffic demands in the AT&T IP Backb
one, by collecting, validating, and joining very large and diverse sets of
usage, configuration, and routing data over extended periods of time. The p
aper concludes with a preliminary analysis of the observed dynamics of the
traffic demands and a discussion of the practical implications for traffic
engineering.