Per Kurose and Ross (2004) the four principles of telecommunications Quality of Service are packet classification, isolation (scheduling and prioritization), high resource utilization and call admission. As these are principles and therefore integral to the definition of QoS there really aren't any circumstances under which they can be completely removed and the remaining system said to supply Quality of Service. However, it certainly is possible that their importance be adjusted so as to deliver a 'best effort' QoS or balanced so that one can have a true QoS that accentuates applicability to the issues of real world scenarios.

Kurose and Ross declare that “packet classification allows a router to distinguish among packets belonging to different classes of traffic.” This means that, in order to supply some Quality of Service over a best effort network some additional layer of information must be introduced. IP, for example, has no mechanism whereby a router can discriminate between 'classes of traffic' (Balliache, 2003). However, putting this information in IP packets and programming routers to recognize and respect it is a trivial matter. Therefore this pillar is achievable and must implemented be in order to have a basic QoS.

“It is desirable to provide a degree of isolation among traffic flows, so that one flow is not adversely affected by another misbehaving flow”, Kurose and Ross go on to say. How integral isolation by this exact definition is to QoS appears debatable. That no flow of traffic be completely starved seems elemental and fair. But, in fact, if one customer is paying for a guaranteed service and one is not, whether the former's application is 'misbehaving' or not is irrelevant to the proper delivery of QoS. However, that the facility to perform this isolation be available in order to speak of Quality of Service is not arguable. Should any customer have a guarantee of some base level of service then the network must be able to isolate flows in order to deliver that. Even if that guarantee doesn't exist the isolation is necessary to control any differentiated flows, however not necessarily per this specific ('misbehaving') example.

Kurose and Ross say of the third principle of QoS that “while providing isolation among flows it is desirable to use resources (e.g. link bandwidth and buffers) as efficiently as possible.” This is an interesting assertion to tie into the second principle. Personally I fail to see why isolation bears on this desirability. It seems to me that efficient utilization of resources is imperative however isolated traffic flows are and therefore the first part of K&R's definition of the third principle is unnecessary. However the core of the principle is anything but, and the very definition of Quality of Service must assume that resources are being used efficiently.

The final principle by which Kurose and Ross define QoS is that of call admission. That a “process is needed in which flows declare their QoS requirements and are then either admitted to the network (at the required QoS) or blocked from the network (if the required QoS can not be provided” seems as integral to the very definition of Quality of Service as any of the other principles. If this blocking process didn't exist then guarantees of quality would be meaningless and/or impossible, as they demonstrate with their example of “Two 1 Mbps Audio Applications over an Overloaded 1.5 Mbps Link.”

It seems to me that Kurose and Ross have defined Quality of Service succinctly and effectively with their four principles. While I think the principles can allow for flexibility and prioritization and still deliver a product we may call Quality of Service, especially the principle of isolation as they state it, I do not believe that any principle can be removed and still have meaningful QoS.

Balliache, K. (2003) Network QoS using Cisco HOWTO – 1.0 Why QoS need be applied? [Online] Available from: (Accessed: 15 November, 2010)

Kurose, J.F. & Ross, K.W. (2004) Computer Networking – A Top Down Approach, 4th Ed. Pearson Education