[mpls] [mpls-tp] Several open issues on draft-xiao-mpls-tp-throughput-estimation-00

[[e-mail]]

Hi all,

In light of the previous comments and discussions on this draft, there are 
still some open issues which need to be closed before updating it, so I 
want to request for your opinions, details as follow:

1. Wouldn't it be better to have a threshold on the acceptable frame loss 
rate and not require absolutely no packet loss?
This question was raised by Ayal Lior ([e-mail]) As I know in 
practice the threshold is a configurable parameter when executing 
throughput measurement, but in RFC1242 and RFC2544 the throughput is 
defined as that way no packet loss permitted.

2. How to align the format of test data packets between 
draft-boutros-mpls-tp-li-lb-01 and this draft?
As I said in my presentation of this draft, IMO in order to simplify the 
implementation, the format of test data packets should be the same whether 
it's used for data plane loopback or throughput estimation. Now the format 
of data packets defined in section 5 of draft-boutros-mpls-tp-li-lb-01 is 
different from the format of test packets defined in section 3.2 of this 
draft. I prefer the format with OAM encapsulation defined in this draft, 
and my reason is that with this format it's easy to distinguish the test 
data packets from other packets when counting them.

3. Is there a need to simplify the procedures defined in this draft and 
how?
I received the comment both in the mailing list and in private discussions 
that the procedures defined in this draft are some complex, but I have no 
idea about the proposed simplified solution, and IMHO the procedures 
defined in this draft are enough efficient and also the usual operations 
for throughput measurement. Hope for your concrete suggestions.

Any other comment and discussion are welcome.

Best Regards,
Xiao Min