QoS: the holy grail - network quality of service - Industry Trend or Event

You've probably heard some heated arguments over network quality of service during the past 12 months, from the trade press and vendors alike. But there hasn't been much talk about user requirements for network service quality level- how various quality of service technologies best serve user applications.

Many people are under the impression that quality of service matters only for future multimedia applications such as real-time video. This is not true. Organizations are increasingly placing multiple types of traffic on the same LANs and TCP/IP backbone networks, causing competition for limited band-width.

Network traffic congestion degrades the response times of critical network applications, a problem that must be dealt with today. The good news is that fixing the network quality of service problem now provides an added benefit--a network infrastructure prepared for emerging real-time voice and video requirements.

It is no longer possible to assume that particular business functions will be carried out in stable ways, in stable locations, or for long periods of time.

Network-connected applications are being combined and integrated for presentation to the user: workstations are beginning to provide messaging, video, telephone, and other work-collaboration functions. The concept of the multi-service network was created to provide an integrated backbone with flexibility and cost savings.

But as these diverse legacy networks with different characteristics and priorities are combined to create that single, enterprise-wide network, we need ways to ensure that network performance is acceptable for each application. If the multi-service network fails the test of user acceptance, special-purpose networks will reappear.

Quality of service technology attempts to provide a method for categorizing and prioritizing traffic to ensure that traffic crucial to the enterprise will always flow across the backbone in a timely manner, regardless of competing demands for bandwidth by less-important applications.

We can group the traffic flowing across an integrated enterprise network into three categories, differentiated by their transit delay needs (see Figure). Within each category, traffic may be further subdivided by priority. Priority traffic receives preferential treatment because of its importance to the enterprise, unlike a traffic category, which is supporting an application that will fail if sufficient communications service is not provided.

Despite the apparent complexity of the choices among quality of service technologies, examination of the technologies and the possible network situations yields only three major options: overwhelming bandwidth, router prioritization, and an ATM core with Ethernet edge switches.

In three to five years, the great majority of enterprise networks will be moving towards integrated network backbones to improve productivity and efficiency. As traffic streams with different characteristics and priorities are combined to create that single, enterprise-wide multi-service network, network performance must still be acceptable for each application.

The network administration must be able to enforce prioritization within switches and for link access, to validate requests for particular service levels, and to justify bills rendered to internal or external customers by verifying delivery of service levels.

The new quality of service application programming interfaces (APIs), which allow users to request quality of service levels dynamically, complicate this problem and reinforce the need for central control to avoid network saturation and the resulting loss of distinction among service classes. The integrated network backbone will need sophisticated, industrial-strength quality of service facilities.

COPYRIGHT 1997 Nelson Publishing
COPYRIGHT 2004 Gale Group