If think of the relation between the network and a user as being defined by a service contract. The contract obligates the network to transfer the user’s information with a defined QoS (delay, loss, rate, etc.) provided that the user’s traffic conforms to specified limits (bit rate, burstiness, etc.). from this view of point, the objective of network control is to fulfill the largest set of contracts.

The Integrated Services require some protocol to communicate with the routers along the path of the intended flow and reserve the resource necessary to provide the service. (A sort of dedicated connection.)

For integrated services, some information has to be communicated between the end systems and the network elements (intermediate points, such as routers and subnets). A logical choice for this communication is RSVP (Resource ReServation Setup Protocol). With functions which deliver the QoS within routers, RSVP provides communications between these functions.

In RSVP, the PATH message flows from the sender to the receiver and contains information about the traffic stream coming from the sender. The traffic stream itself is described by the Tspec of the sender, which contains ADSPEC and T(raffic) Spec objects. The Tspec signals to all intermediate nodes and the destination what the characteristics of the stream are (data information). The ADSPEC can be originated by the sending host or the first router in the path, and contains a number of general parameters characterizing the path which are determined by hop-by-hop computations, and also parameters characterizing particular QoS services supported by the network elements (path information).

This service is for a certain class of applications which have a ‘real time’ character, but can adapt somewhat to network conditions, but which tend to perform badly on loaded networks. It dosen’t provide service guarantees.

It addresses specifically two QoS parameters: packet loss, and delay. The goals of this service are that the packet loss rate "closely approximates" that of the medium and that the transit delay through network elements is close to the minimum possible transit delay (i.e. no queues).

The figure shows a network element which is providing controlled load service. The packet classifier and the token bucket are for classifying all input packets. Packets which conform to the token bucket are put into the queue of priority. Packets which do not conform to the token bucket and packets which receive only best-effort service are put into the other queue.

The purpose of guaranteed service is to provide bounded delay and assured delivery of all packets for some kind of service. An essential part of guaranteed service is that the parameters from all the network elements in a particular path have to be able to be combined in a simple way to predict (bounds on) the QoS that will be experienced by the packets in that flow.

Since guaranteed service delivers specific delay bounds, both the source and the network elements have to give specific behavior.

So a token bucket envelopes the source, so the source stream must at all time stay within the bounds set up by the token bucket.

It is a weaker method of distinguishing different flows of packets and requires some policing, but dose not necessarily require the state information associated with the flows defined by integrated services.

Among the mechanisms discussed for QoS Support, the most elementary are:

Priority queuing: today’s packet networks are truly democratic and egalitarian. FIFO is about the only packet forwarding strategy. Priority queuing could alleviate jitter problems.

Custom queuing: this is a bandwidth reservation mechanism. It is less drastic than priority queuing as it schedules MM packets based on the specified jitter upper bound, not on the absolute priority.

Two possible solutions:

ATM networking technology – QoS is a build-in feature for this type of transport. This is probably a solution for tomorrow’s GII (Global Information Infrastructure ) backbone. Will it work? 100% ATM network will never happen. ATM far more expensive than broadcast networks; new fiber/UTP cable plant needed, Serious management and maintenance problems unsolved. ATM QoS does not provided global QoS support.

Problems: Solution: Integrated services over the packet switched networks. A new architecture on top of the current IP layer as opposed to ATM low level approach.

The integrated services infrastructure complements two other enabling technologies:

Client multimedia hardware:

IP multicasting: integrated services architecture assumes multicast as a sine quanon (sanguineous) component of future multimedia Internet. (Actually, real-time QoS is not the only issue for a next generation of traffic management in the Internet: network operators are requesting the ability to control the sharing of bandwidth on a particular link among different traffic classes. They want to be able to divide traffic into a few administrative classes and assign to each a minimum percentage of the link bandwidth under conditions of overload, while allowing "unused" bandwidth to be available at other times. These network administration classes may represent different user groups or different protocol families, say, controlled link-sharing. Therefore, the integrated services (IS) for an Internet service model includes best-effort service, real-time service, and controlled link sharing.)

The IS model includes two sorts of service targeted towards real-time traffic: guaranteed and predictive service. There are several basic assumptions of the model: resources (e.g., bandwidth) must be explicitly managed in order to meet application requirements. The IS model proposes to use the existing Internet-layer protocol for real-time data.

The implementation framework includes four components:

The packet scheduler,

The admission control routine,

The classifier,

And the reservation setup protocol.

QoS controller: functions in ATM switches. Negotiates with service requester, and allocates resources.

Info collector: functions in the switches, and client-side work stations. Collects delay, jitter statistic information from the switches and the clients.