At 08:47 AM 05/22/97 -0700, Mike Trest wrote:
I hope that Paul's collections of comments (public and private) will be summarized for this community. However, it would be interesting to see the discussion broadly published in Internet trade press with a clear statement of the issues with a WWW mechanism to reply with a yes-mabey-no statement of interest.
I encourage this consumer polling because I do not assume opinions of NANOG readers are representative of opinions in the business market.
Some of the responses I received on this issue were copied to the list, others were not. I've summarized the responses below, and taken the liberty of boiling down the results. I'm not really enthusiastic about the volume of responses I received, but I suppose it's better than none. What this indicates to me is that the majority of you have no opinion on the matter of QoS or are simply not interested. I will probably post the same questions to the cisco@spot mailing list [USENET: comp.dcom.sys.cisco] to get a representative sampling of non-Internet -related (corporate) perspectives. Enjoy. - paul [snip] Number of respondents: 19 Number of respondents who felt that better QoS 'knobs' were needed in the routers: 4 Number of respondents who felt that admission control and policing functionality was required: 2 Number of respondents who feel that QoS granularity at the IP source/destination and/or tcp/udp port level is sufficient: 2 Number of respondents that indicated that QoS should have the granularity to differentiate with per-flow granularity: 1 Number of respondents who need QoS differentiation for enhanced economic factors (charge more money): 4 Number of respondents who indicated that (paraphrased) QoS isn't interesting due to over-engineering: 2 Number of respondents who indicated that (paraphrased) QoS would indeed be interesting if they were congested: 1 Number of respondents that indicated that congestion management needs to be an integral part of any QoS implementation: 1 Number of respondents who indicated that until QoS routing was available, QoS was not interesting: 1 Number of respondents who indicated that inter-domain QoS transit was a major stumbling block: 6 Number of respondents that explicitly mentioned RSVP by name: 3 Number of respondents that explicitly mentioned RSVP in a favorable tone: 2 Number of respondents who indicated that QoS implementations, whatever they may be, needs to be pervasive, to include hosts: 2 Number of respondents who indicated that billing & accounting systems need to be developed for QoS: 4 Number of respondents who indicated that the term 'QoS' is too ambiguous: 2 Number of respondents who indicated they expect guaranteed delivery in a QoS implementation: 0 [snip]