Hi Tom, Many thanks for your kindly help. Please see inline.
Best regards, Bo > -----邮件原件----- > 发件人: tom petch [mailto:[email protected]] > 发送时间: 2022年1月26日 1:09 > 收件人: Wubo (lana) <[email protected]>; Wubo (lana) > <[email protected]>; [email protected]; > [email protected] > 主题: Re: [OPSAWG] I-D Action: draft-ietf-opsawg-yang-vpn-service-pm-02.txt > > From: Wubo (lana) <[email protected]> > Sent: 25 January 2022 13:16 > > Hi Tom, > > Thanks for the review and we will correct these issues in the next version. > Please see inline. > > <<tp> inline twice> > > Best regards, > Bo > > > -----邮件原件----- > > 发件人: tom petch [mailto:[email protected]] > > 发送时间: 2022年1月21日 20:30 > > > > From: OPSAWG <[email protected]> on behalf of Wubo (lana) > > <[email protected]> > > Sent: 20 January 2022 03:24 > > > > > > percentile definition seems not quite right. You are using a > > percentage here but there is no requirement to do so - it can be an > > integer count or any other measurement of a proportion > [Bo Wu] We provide two options here, and some providers think percentile is > also useful. I'm not sure the issue. Could you elaborate a little bit? > > <tp> > Percentile is fine as a way of characterising a distribution with a single > value. > My comment is that for any percentile, be it 50-percentile, 95-percentile or > whatever, then there is a value to go with that and that value can be anything > that is a value - integer, percentage etc. I think that the description here > wants it to be a percentage which I think too limiting. > [Bo Wu] Thanks for the explanation. How about the following change? OLD: typedef percentile { type decimal64 { fraction-digits 5; range "1..100"; } description "The percentile is a statistical value that indicates that a certain percentage of a set of data falls below it."; } NEW: typedef percentile { type decimal64 { fraction-digits 2; range "0..100"; } description "The percentile is a value between 0 and 100, e.g. 10, 99.9 ,99.99 etc. . For example, if the percentile is set to 95. For a given one-way delay measurement, if the 95th percentile one-way delay is 2 milliseconds, then the 95 percent of the sample value is less than or equal to 2 milliseconds."; } > > > > 'middle-delay' I think confusing; sounds too much like it is the > > median or mode; perhaps medium? > [Bo Wu] Agree. rfc3393 uses median to indicate 50th percentile. so we change > it to "median-delay"? > > <tp> > Well no, it is median only when it is a 50-percentile which is only the > default > here and if the user chose 90 instead, wanting the three percentiles to be 60, > 90, 95, then I think median is misleading. > > And RFC3393 says (along with defining ipdv:-) that the 50-percentile is the > median which is correct. Here you are defining low, something else, high and > when something else is not 50, then it is not the median. I am struggling > for a > word that sits between low and high that does not have connotations of being > the middle! intermediate? [Bo Wu]Got it. Thanks. We will change it to "intermediate" to avoid the confusion. _______________________________________________ OPSAWG mailing list [email protected] https://www.ietf.org/mailman/listinfo/opsawg
