Re: draft-ietf-bfd-base-03.txt comments

[Chezhian Renganathan <[e-mail]>]

Dave Katz wrote:

>
> On Aug 17, 2005, at 10:44 AM, Chezhian Renganathan wrote:
>
>> Hello,
>>
>> Section 6.7.4 -
>> Unless I am missing something, it seems to me that the statement in  
>> braces
>> "roughly speaking, due to jitter"
>> should be applicable in the context of Detection Time rather than  
>> DetectMult.
>> The DetectMult is just number of packets and there is no variance  
>> there with
>> jitter.
>
>
> The text in the document is correct.  DetectMult is used to calculate  
> time, not count packets, so given the application of jitter, the  
> number of packets missed is approximate (if DetectMult were 10, for  
> example, the actual number of packets missed before detection might  
> be as high as 13 if 25% jitter is applied and the timing is right.)
>
Ok got it.

>>
>> Also, was there any thinking along the lines of making the "Reg Min  
>> Echo RX
>> Interval" field optional, keeping with the spirit of this function,  
>> being optional
>> with either modes. This field could potentially be set to all zeros  
>> for the entire
>> length of a session unless changed if the end systems support this  
>> function.
>
>
> Not sure exactly what you're saying here.  To actually leave out the  
> field and shorten the packet four bytes?  I'm not sure what would be  
> gained by this.  The processing cost would be higher (Is the field  
> there?  What is the value?) and the difference in packet size  
> negligible.
>
Would it help the processing cost if done similar to the auth field 
which is controlled 
by the "A" flag.?
Apart from reducing the overhead there is some bandwidth being saved 
too. For instance, when
asynchronous mode is active across end systems (and potentially multiple 
sessions),
with no echo support, with a 10 msecs tx interval, we could be sending 
these 4 bytes
unnecessarily over the period of the session(s) uptime which could run 
into several days
 between interruptions.

thanks,
chez

> As it stands now, the field can be set to zero if the sender does not  
> wish or is not able to support the echo function.
>
> Perhaps I don't understand the question.
>
> --Dave
>
>>
>> thanks,
>> chez
>>
>>
>>
>
>