ity. This may also initiate a network planning in
order to improve network design, capacity, tech-
nology, and element configuration.
The actual relations between the process model
and the context are not elaborated on in
[ID_tepri]. However, considering that an early
phase is to assess conditions of the network,
most of the process model relates to an opera-
tional network. The context/setting would also
include perspectives on a longer time, e.g. allow-
ing for more aspects to be considered.
3.3 TE Key Components
The key components of the TE process model
are (see Figure 7):
- Measurementsubsystem: Carrying out mea-
surement is essential to providing feedback
on the system state and performance. It is also
critical in order to assess the service level
provided (and QoS) and effect of TE actions.
A basic distinction between monitoring and
evaluation is to be observed; monitoring refers
to the provision of raw data, while evaluation
refers to the use of the raw data for inferring
on the system state and performance. Measure-
ments can be carried out at different levels of
aggregation, e.g. packet level, flow level, user
level, traffic aggregate level, component level,
network-wide level, and so forth. In order to
perform measurements systematically, several
questions have to be answered, like [ID_tepri]:
Which parameters are to be measured? How
should the measurements be accomplished?
Where should the measurements be per-
formed? When should the measurements be
performed? How frequently should the moni-
tored variables be measured? What level of
measurement accuracy and reliability is desir-
able and realistic? To what extent can the mea-
surement system permissibly interfere with the
operational network conditions? What is the
acceptable cost of measurements? - Modelling and analysissubsystem: A central
part of the modelling is to elaborate a repre-
sentation of the relevant traffic characteristics
and network behaviour. In case a structural
model is used, the organisation of the network
and its components are the main emphasis.
When behavioural models are used, the
dynamics of the network and traffic are the
key issues. The latter model is particularly
relevant when performance studies are under-
taken. Then adequate models of the traffic
sources are also needed. - Optimisationsubsystem: Optimisation can be
categorised as real-time and non-real-time.
The former operates on short to medium time
scales (e.g. ms to hours) and works on adjust-
ing parameters in mechanisms in order to
relieve congestion and improve performance.
Examples of means are tuning of routing
parameters, tuning of buffer management
mechanisms and changing Label Switched
Paths (LSPs). Non-real-time is also seen as
network planning, typically working on a
longer scale. For both of these, stability and
robustness are essential concerns.Routing is a central component in efficient han-
dling of traffic flows in an IP-based network.
When introducing a number of service classes,
some additional constraints can also be consid-
ered when deciding upon the possible routing.
Examples of such constraints are available band-
width, hop count, and delay. This implies that
possible paths as seen from a router must have
the corresponding attributes attached.3.4 Mechanisms and Subjects
In order to complement the best effort service, a
number of activities are undertaken by different
IETF groups as well as by others. The subjects
listed below are treated more in detail in accom-
panying papers of this Telektronikkissue:- Integrated Services (IntServ). Applying this
service model requires that resources are
reserved before the traffic flow starts. As men-
tioned earlier, transmission links and buffers
are commonly seen as resources. Mechanisms
like packet classifiers, packet schedulers and
admission control units have to be present in
the routers supporting IntServ. A classifier
identifies flows that are to be served with a
certain level. A scheduler handles the service
scheduling to ensure that requirements of the
traffic flow are met. Admission control deter-
mines whether or not a router has the needed
resources available to accept a new flow while
still meeting all requirements for all flows pre-
sent. Two additional service classes are identi-
fied: guaranteed service and controlled-load
service. As state information has to be kept for
each group of traffic flows, a router in a larger
network may have capacity problems keeping
all that information. Hence, IntServ is fre-
quently claimed to face the scalability prob-
Figure 7 TE key components/
subsystemsOptimisation
subsystem:- real time and
Modelling andanalysing non-real time
subsystem: - traffic
- network
- resources
Measurement
subsystem:- evaluation
- monitoring