We consider routing schemes for connections with end-to-end delay requireme
nts, and investigate several fundamental problems. First, we focus on netwo
rks which employ rate-based schedulers and, hence, map delay guarantees int
o nodal rate guarantees, as done,vith the Guaranteed Service class proposed
for the Internet. We consider first the basic problem of identifying a fea
sible route for the connection, for which a straightforward yet computation
ally costly solution exists. Accordingly, we establish several approximatio
n schemes that offer substantially lower computational complexity. We then
consider the more general problem of optimizing the route choice in terms o
f balancing loads and accommodating multiple connections, for which we form
ulate and validate several optimal algorithms. We discuss the implementatio
n of such schemes in the context of Link-state and distance-vector protocol
s. Next, we consider the fundamental problem of constrained path optimizati
on. This problem, typical of quality of service routing, is NP-hard. While
standard approximation methods exist, their complexity may often be prohibi
tive in terms of scalability. Such approximations do not make use of the pa
rticular properties of large-scale networks, such as the fact that the path
selection process is typically presented with a hierarchical, aggregated t
opology. By exploiting the structure of such topologies, we obtain an E-opt
imal algorithm for the constrained shortest-path problem, which offers a su
bstantial improvement in terms of scalability.