Illustrated TCP/IP Illustrated TCP/IP
by Matthew G. Naugle
Wiley Computer Publishing, John Wiley & Sons, Inc.
ISBN: 0471196568   Pub Date: 11/01/98
  

Previous Table of Contents Next


Chapter 324
Resource Reservation Protocol (RSVP)

QoS (Quality of Service) is currently limited to manual items such as filters, protocol prioritization (fancy filters), compression, network design, and fat pipes. Most of these techniques are applied to WAN ports. While these interim solutions work well, many applications such as voice and video are running on LANs and WANs. There is another step to providing QoS with broadcast networks—RSVP. RSVP provides a general facility for creating and maintaining distributed reservation states across unicast or multicast environments. It is not supposed to be the QoS that rivals ATM’s guaranteed QoS. It shows promise as the first of many entries into building QoS for existing broadcast-oriented networks without having to tear out a network and replace it with ATM.

Quality of Service has never been built into most protocols that are currently running on networks today. When Ethernet was invented in the late 1970s, 10 Megabits seemed a huge-enough pipe to give any bandwidth-hungry application more than enough room. However, a bandwidth-hungry application is not the culprit; the culprit is millions of bandwidth-hungry users. Personalizing the computer was not thought to have a great impact on the business world. Mainframes and minicomputers were expected to continue to be the computing source of choice; however, the PC changed that. After a few years, the personal computer became able to handle sophisticated graphics, and many different options of voice and video soon became available. Connection to the Internet became a must-have as well.

Resource Reservation Protocol (RSVP)

  QoS abilities on most IP networks are usually about filters, protocol prioritization (fancy filters), compression, and fat pipes (fast or gigabit Ethernet).
  QoS never seemed like a pressing issue until the Web.
  Cannot continue to simply provide for fatter pipes.
  We must find a way to allow multimedia to work on the existing infrastructure.
  ATM is not an alternative for most implementations.

Shared Ethernet and Token Ring networks could not provide the bandwidth necessary to support not only bandwidth-intensive applications that are network aware, but the millions of personal computer users as well. Ethernet has since scaled to 100 Megabits per second and Gigabit Ethernet is making inroads as well. The virtually limitless scalability of the ATM protocol is the first commercial protocol that has QoS scaleable parameters built in; however, ATM is still less than 1 percent of all desktop installations. And there are many consumers that will not tear down their Ethernet or Token Ring networks and replace it with ATM just to get QoS and scaleable bandwidth. Consumers want QoS, but they want it with their existing networks.


Previous Table of Contents Next