differently by different networks along a path,
leading to a vast number of options. Harmon-
ising the service levels may result in easier
interconnection and service provision in mul-
tiprovider environments, although service lev-
els could also be seen as a competitive factor.
Service discovery as mentioned above be-
comes even more requested for such configu-
rations. Between the different networks this
could also ask for enhancing the set of routing
protocol attributes, including some describing
the service levels supported.
- Measuring service level and delivery. In sell-
ing a service level to a customer, it is central
to have means in place in order to document
that the service has been delivered as agreed.
Another purpose of such measurements could
be as a basis for admission control and routing. - Accounting for service levels. In case various
service levels are to be offered, corresponding
range of tariffing levels should be used as
well. A technical argument is that use of more
network resources should be reflected in a
higher tariff. Naturally, other arguments may
go against such a conclusion. This technical
argument points towards application of usage-
based charging.
According to [RFC2990] the following aspects
are included in an architecture for service level/
QoS level:
a) Control the network response such that it is
consistent and predictable;
b) Control the network response such that the
service level is provided as agreed:
c) Allow establishment of agreed service level in
advance;
d) Control contention for network resources such
that the appropriate service levels are achieved;
e) Control contention for network resources such
that a fair allocation is achieved (although fair
has not been defined);
f) Allow for efficient utilisation of network
resources while providing a range of service
levels.
All these issues have to be addressed in order to
ensure that the service levels can be provided.
Actually, all these have to be in place in a coher-
ent way to offer end-to-end service levels in
agreed ways.
In addition to the issues listed above, more unan-
swered questions are found for interdomain con-
figurations, like how to efficiently handle the set
of SLAs in a multi-service and multi-provider
configuration.
6.3 Overview of Further Challenges
In addition to the issues discussed above, more
aspects can be looked at. The following descrip-
tion of the open issues is centred around the
groups depicted in Figure 9:
- The network nodes sphere representing vari-
ous nodes involved for IP transport, e.g.
routers and hosts/terminals. Various segments
of an operator’s network, including access and
core, are part of this. User terminals and rele-
vant parts of applications can also be included.
Typical issues are related to ensuring and
monitoring performance of traffic flows, con-
figuring resources, and so forth.
Figure 9 A grouping of
challenges related to ensuring
QoS for IP-based services
Regulator, vendors
User - Provider Service
Level Agreement
Multiple providers
may be involved
User
business
Provider
business
Actual delivery
Service, service control, service management etc.
IP transport, network nodes, network management etc.