3.1 Definition of Constraints
Constraints and resources are counterparts to one
another: routes have constraints while network
elements (nodes and links) have resources. As
paths are explored, the constraints for a route are
checked against the resources along the path to
see that the constraints are met. The constraints
specified must match available resource infor-
mation [CR-notes].
Constraints can be divided into Boolean and
quantitative. Some constraints can be of both
types. Boolean constraints indicate whether or
not a candidate path is feasible. Quantitative con-
straints assign numerical values to paths, en-
abling choice between feasible candidate paths.
Resources can be divided into configurable,
dynamic and topological. Configurable
resources are those assigned by an administrator,
e.g. administrative groups and link metrics.
Dynamic resources are those that depend on the
network state and vary with time, e.g. available
link bandwidth. Topological resources are those
that are enforced by the topology of the network,
e.g. path length.
3.1.1 Boolean Constraints
Boolean constraints include (related resource in
brackets):
- Administrative group constraints (administra-
tive groups or colours configured on links); - Bandwidth availability (available link band-
width); - Delay bounds (configured delay on links and
nodes); - Hop count bounds (path length).
3.1.2 Quantitative Constraints
Quantitative constraints include (related
resources in brackets):
- Residual bandwidth ratio (residual link band-
width); - Path metric (metric);
- Resilience (penalty, applies to backup path
computation); - Hop count (path length).
In order to select a path among feasible candi-
date paths, the quantitative constraints have to be
ordered or prioritised in some way. This order
should be administratively configurable. A
default ordering of the quantitative constraints
could e.g. be: path metric, resilience, residual
bandwidth ratio and hop count (suggested by
[CR-notes]).
3.2 QoS Routing
A definition of QoS(-based) routing [RFC2386]:
A routing mechanism under which paths for
flows are determined based on some knowledge
of resource availability in the network as well as
the QoS requirement of flows.Another definition of QoS routing [QoSGlos]:
A dynamic routing protocol that has expanded
its path-selection criteria to include QoS parame-
ters such as available bandwidth, link and end-
to-end path utilisation, node resource consump-
tion, delay and latency, and induced jitter.QoS routing is regarded as a subset of the more
general constraint-based routing concept. It
selects routes with sufficient resources for re-
quested QoS parameters.The main objectives of QoS routing are:- Dynamic determination of feasible paths;
- Optimisation of resource usage.
The QoS requirement of a flow is given as a set
of constraints, which can be link constraints,
path constraintsor tree constraints(applicable
to multicast flows only). A link constraint speci-
fies a restriction on the use of links. A band-
width constraint of a unicast path will e.g.
require that the links constituting the path must
have a minimum amount of available bandwidth.
A path constraint specifies the end-to-end QoS
requirement on a given (single) path, while a
tree constraint specifies the QoS requirement
for the whole multicast tree. A feasible path is
a path that has sufficient unused resources to
satisfy the QoS constraints of a flow. The basic
function of QoS routing is to find such a path.
Additionally, the applied QoS routing algorithm
may try to optimise resource utilisation by con-
sidering link cost. The optimal output of a QoS
routing procedure is the lowest-cost path among
all feasible paths.3.2.1 Network State Information
In order to find a feasible path for a new flow, it
is necessary to have up-to-date state information.
The state information may be classified as de-
scribed in the following.Each node is assumed to maintain its up-to-date
local state, including the queuing and propaga-
tion delay, the residual bandwidth of the outgo-
ing links, and the availability of other resources.The combination of the local state of all nodes in
the network is called a global state. An IGP with
appropriate TE extensions may be used to spread
this information among the network nodes so
that each node knows the topology of the net-
work and the state of each link as illustrated in
Figure 3.1. The information kept by the nodes