Quality-of-service issues are no longer quite being ignored, or, at least, not blithely. Streaming video is here to stay on a large scale. Nonetheless, VoIP at least is here to stay on a medium scale. it is not clear how to compare special traffic like multicast, for the purpose of pricing, with standard unicast traffic.it is not clear how to charge endpoints for their QoS requests (as with RSVP), particularly when the endpoints are not direct customers.However, often an ISP’s problem with a QoS feature comes down costs: routers will have more state to manage and more work to do, and this will require upgrades. Sometimes this is simply the chicken-and-egg problem: ISPs do not like to implement features nobody is using, but often nobody is using them because their ISPs don’t support them. We introduce the theory, anyway some of these mechanisms have not exactly been adopted with warm arms by the ISP industry. In this chapter, we introduce some quality-of-service mechanisms for the Internet. This is a major departure from the classic Internet model of “stateless” routers that have no information about specific connections or flows, though it is a natural feature of virtual-circuit routing. These routers will then use the queuing and scheduling mechanisms of 23 Queuing and Scheduling to set aside bandwidth for designated traffic. In any network, large or small, in which bulk traffic may sometimes create queue backlogs large enough to cause unacceptable delay, quality-of-service assurances must involve the cooperation of the routers. Furthermore, real-time applications may simply fail – at least temporarily – if these bandwidth, delay and loss constraints are not met. Real-time senders are likely to have not only bandwidth constraints, but constraints on delay and on loss rates as well. While any sender might be interested in quality-of-service levels, they are an especially common concern for those sending and receiving real-time traffic such as voice-over-IP or videoconferencing. Even bulk senders, for example, might sometimes wish to negotiate ahead of time for a specified amount of bandwidth. Such arrangements are known as quality of service (QoS) assurances, and may involve bandwidth, delay, loss rates, or any combination of these. We may allow some individual senders to negotiate such services in advance, or we may grant preferential service to specific protocols (such as VoIP). Sometimes, however, we may want some traffic to receive a certain minimum level of network services. For bulk file-transfers this is usually quite sufficient one way to look at TCP congestive losses, after all, is as part of a mechanism to ensure optimum utilization of the available bandwidth. Delays and losses due to congestion are nearly universal. So far, the Internet has been presented as a place where all traffic is sent on a best-effort basis and routers handle all traffic on an equal footing indeed, this is often seen as a fundamental aspect of the IP layer.
0 Comments
Leave a Reply. |