Hi, Sabine:
As we know, link specific metric can be aggregated into a new network metric, 
let's call it path metric or end to end metric.
I am wondering whether we should introduce a new metric or we stick to the 
existing metric we defined in draft-ietf-alto-performance-metrics-01.
Take "availbw" as example, availbw is defined as TE specific metric or link 
specific metric(i.e., metric is measured on the link, if we want to introduce 
another metric to measure available bandwidth on the path from source to 
destination, we have two to do that:

1.       Define "path-availbw" metric, the measurement methodology for 
"path-availbw" is similar to one we used for link-availbw

2.       Stick to "availbw"metric we define in the document, but we use another 
parameter, e.g. using context-parameter to distinguish link availbw from path 
availbw. In this cse, the measurement methodology is same.
Another example is in draft-ietf-alto-performance-metrics-01, we define one way 
delay and round trip delay, there are two different cost metrics and 
measurement methodology for these two are different. I am wondering whether we 
should have a single metric delay metric and use some other parameter to 
distinguish one from another? Yes, context-parameter add a lot of flexibility 
and but also add complexity. How do you make sure the cost metric with 
different context can be parsed correctly.
What do you think of this? Comments and Suggestions?

-Qin
_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto

Reply via email to