Side_1_360

the end-hosts or a Service Provider (SP). The

statements handling the performance and the

service level should be expressed in terms of

application units that the user understands and

is concerned about, e.g. the time to complete

the transaction instead of round trip delay and

some additional information. This type of

agreement also includes the constraints on the

user and his behaviour and minimal require-

ments placed on the equipment, since the

application using network services is known.

It is a common practice today that the charac-

teristics of the customer owned/administered

equipment (i.e. user’s network) are described

by using agreed parameters. When consider-

ing the QoS-related part, a selection of QoS

parameters must be devised by focusing on

the application, e.g. a parameter of availability

of 99.995 % of time may be expressed as

the only 25 minutes per year (3 minutes per

month) that the service will be unavailable.

The values of the relevant QoS parameters are

determined by taking into account the pecu-

liarities of a particular implementation of the

service.

2.Network-level SLA– specifies statements in
terms of performance observed while provid-
ing a network connectivity / transport service.
This agreement is usually made between two
providers, though in case a customer is e.g.
a large company it may be made between an
end-customer and a provider agreeing upon
the usage/provision of a transport service.
One type of the network level SLA is a peer-
to-peer agreement made between ISPs. The
parameters used to describe the performance
of the network and the quality of the network
service(s) are very detailed, technical parame-

ters whose objectives may be described by

using different statistics and moments.

Depending on the scope of the service, and its

implementation various types of network level

SLAs exist. For example, a Leased Line (LL)

service may be supported by the infrastructure

that is Asynchronous Transfer Mode (ATM)

based or Frame Relay (FR) based, for which

ATM- and FR-related parameters are used,

respectively. The network level SLAs for IP

transport services are elaborated on in Chapter

4.

3.Service provider SLA– used by providers

offering server-hosting capabilities. The

provider (usually an actor playing a role of a

service provider, or a co-location provider)

has control over the server side but not over

the customer side or the network performance.

In this case, parameters related to the perfor-

mance of (various types of) servers, or other

(peripheral) equipment are relevant. The

parameters may include the performance of a

database, e.g. a number of simultaneous trans-

actions that a database can serve, a number of

simultaneous hits on the web-content that a

web server can support, etc.

3.2 Generic Structure of the

QoS Part in an SLA

In order to handle both the increasing volume of

SLAs, their complexity and to assure their main-

tenance, having the generic structure/template

that can always be (re)used would help a lot.

Being genericimplies the structure’s indepen-

dence of service type, network, technology

involved in service provisioning, type and

organisation of actors involved, etc. On the other

hand, being generic does not exclude considera-

user SLA provider

Service

description

Legal

issues

Economy/

prices Regulatory Others

Service Level Agreement

Interface

description

Traffic

pattern

QoS parameters

and objectives

Measurement

schemes

Reaction

pattern

Quality of

Service

service

Figure 5 A structure of an

SLA – focus on the QoS part