Two types of QoS architectures are currently available: Integrated Services (IntServ) and Differentiated Services (DiffServ). IntServ is designed to provide guaranteed resources on the network for applications. For example, a network application via Resource Reservation Protocol (RSVP), a signaling protocol, requests specific bandwidth from the network for a given flow. RSVP visiting each network hop toward the IP destination address will attempt to allocate network bandwidth resources. If the results are positive, the application can send its traffic on the network. RSVP in addition to bandwidth allotment can also help allocate resources for delay-sensitive applications as well as provide the traditional best-effort service. The idea behind IntServ is that if specific application traffic flow requirements are met, the likelihood of network affecting the performance of the application is minimized. IntServ requires all devices along the path of the traffic flow to support RSVP. Most networks today cannot meet this requirement.
The second type of QoS architecture is DiffServ. Essentially, some user-defined traffic will have a higher priority than other traffic on the network. For example, the network engineer might provide more network resources for voice traffic over data traffic, and so on. Unlike IntServ, DiffServ is not an all-or-none mechanism. No admission controls mechanism exists in DiffServ as in IntServ. Each network device along the traffic flow will appropriately handle the traffic based on that particular device configuration. Most networks today operate using the DiffServ model because it is more scalable and flexible than IntServ. The discussion of this chapter will be based on DiffServ model.