Network traffic consists of a number of streams generated by applications of end stations. These applications differ from each other with different maintenance requirements and network performance. In fact, the requirement for the maintenance of each flow is completely determined by the requirements of the application that generated this is flow. Therefore, in order to clarify the structure of existing service requests in the network, it is necessary to determine the types of network applications.
The ability of the network to provide the different levels of service requested by one or another network application, along with performance monitoring – bandwidth, delay/jitter and packages loss – can be classified into three categories listed below.
• Nonguaranteed data delivery (best-effort service). Ensuring connectivity of network nodes without the guarantee of time and the fact of delivery of the package to the point of destination. It should be noted that discarding a packet can occur only in the buffer overflow by the input or output queue of the router.
In fact, non-guaranteed delivery of packages is not part of QoS due to lack of quality of service and packages’ delivery guarantee. It should be noted that non-guaranteed delivery of packages is currently the only service supported on the Internet. Despite some performance degradation, for most data-oriented applications (for example, applications that support file transfer protocol (File Transfer Protocol – FTP)), this service is quite sufficient. In general, the optimal operating conditions of all applications include the requirements for allocating certain network resources in terms of bandwidth, delay and the level of packet loss.
• Differentiated service. Differentiation of service involves the separation of traffic into classes based on the requirements for quality of service. Each traffic class is differentiated and it is processed by the network in accordance with the QoS mechanisms specified for this class. Such a quality of service (QoS) scheme is quite often called the CoS scheme. It should be noted that differentiated services themselves do not intent the provision of guarantees for the services provided. In accordance with this, the traffic is divided into classes, each of which has its own priority. For this reason, differentiated services are often called soft QoS (soft QoS). Differentiated service is convenient to use in networks with intensive application traffic. In this case it is important to ensure separation of the administrative traffic of the network from all other traffic and assign it a priority, at any time, be sure of the connectivity of the nodes of the network.
• Guaranteed service. Guaranteed maintenance involves redundancy of network resources in order to meet specific maintenance requirements of the traffic flows. In accordance with guaranteed service, a preliminary back-up of network resources is made along the whole traffic trajectory. Guaranteed maintenance is often called still hard QoS (hard QoS) in connection with the presentation of strict requirements to the network resources. Unfortunately, reservation of resources along the entire path of individual traffic flows cannot be done on the scale of the Internet backbone, serving thousands of data streams at a particular time. To correct the situation, an aggregated reservation of resources is called for, requiring storage in the basic Internet routers are just a small amount of information.
Applications requiring guaranteed service include multimedia applications that transmit voice and video. Interactive applications, focused on the transmission of speech by Internet, can function normally (i.e. without causing inconvenience to the users) only if the latency value is equal to or less than 100 ms.
It should be noted that a similar level of latency is acceptable for most multimedia applications. But the applications of Internet telephony already need an information transfer channel with a bandwidth of at least 8 Kbit / s and a receive acknowledgment delay value of 100 ms. In order to satisfy such requirements for the guaranteed service, the network must have a certain amount of resources.
The level 2 service quality of the OSI reference model (Layer 2 QoS) includes all the QoS mechanisms provided by different link-layer technologies or the technologies, the object of which this level is.