At 3:25 AM +0000 7/20/03, " Chuck Whose Road is Ever Shorter " wrote: >""Howard C. Berkowitz"" wrote in message >news:[EMAIL PROTECTED] > >> What's the medium cost between the two cities? Can you use demand >> circuits as a backup? Can you live with one more PVC and trust the >> physical connection? Is QoS-unpredictable cable or DSL available? >> > >Funny you should ask this, Howard. I've been struggling for several weeks >how to pose the question. Have we, the engineering / technical sales >community oversold the idea of dedicated bandwidth and QoS?
Perhaps not sufficiently refined the ideas. There is a common assumption that "dedicated bandwidth" means "large amounts of dedicated bandwidth." But for IP telephony, the issue is more having quick and reliable access to a small amount of bandwidth. Remember 802.4 Token Bus, which people (especially General Motors) was essential for process control? It was eventually superceded by carefully tuned 802.3, kept lightly loaded and with small MTUs to encourage interleaving. I deal with medical imaging applications where workstation access at 100 Mbps or more is a valid requirement. Still, the real requirement is often dedicated bandwidth from the desktop to a workgroup disk cache, not network-wide to the cache when it may be several days before the radiologist reads the image. There are applications where the end-to-end bandwidth makes sense. I know of a US radiologist that has made a business of providing night interpretation services for smaller Australian hospitals, and has to get the image in minutes. When one considers the people cost here, transoceanic bandwidth begins to look attractive. Things like telepresence surgery, of course, MUST have dedicated bandwidth. Yet I'm also working with battlefield medical systems where I have to design for the communications links being down, or only providing 2.4 Kbps on the WAN. Clearly, this doesn't fly for telepresence, but still has value for logistics. > >You mention QoS in your response above. QoS is something being pushed as >necessary for voice, video, and other delay sensitive traffic. Cisco >wireless AP's offer one way quasi QoS. Wireless, however, remains a >contention medium, and will remain so until the FCC changes the rules. I'm >not sure they will be able to release sufficient radio spectrum to permit >all the bandwidth and services that wired can. But wireless is so damn >convenient! One of the arguments for QoS features is that it doesn't need a highly skilled capacity planner/performance analyst. True, but there may be alternatives, such as more use of expert systems for capacity planning and less real-time QoS > >I'm not suggesting that dedicated bandwidth to the desktop is a bad thing or >that there is not need for QoS. However, I'm wondering how all of us might >reconcile two seemingly opposed points of view regarding bandwidth and QoS - >recognizing that wireless, whatever it's limitations, is here to stay, and >will become and remain essential to any and all networks, enterprise or >small business, going forward. They are all tools in a toolbox. I am now dealing with applications where wireless simply is not acceptable, either for battlefield or hospital use. Remember that simply being able to locate a military transmitter is sufficient to target it--cryptography doesn't help. In hospitals, you frequently have shielded rooms that block wireless propagation. While the hospital-wide bans on cellular telephones and the like may be overreaction, there are very real reasons to be extremely careful of any free-space electromagnetic energy in areas where electrodes penetrate the skin and provide low-resistance paths to the heart. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=72655&t=72645 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
