Re: [Spud] FW: New Version Notification for draft-hildebrand-spud-prototype-02.txt

[Brian Trammell <[e-mail]>]

Sent from my iPhone (but not top quoted)

> On 03 Mar 2015, at 19:50, Ted Hardie <[e-mail]> wrote:
> 
>> On Tue, Mar 3, 2015 at 10:30 AM, Joe Hildebrand (jhildebr) <[e-mail]> wrote:
>> On 3/3/15, 7:28 PM, "Ted Hardie" <[e-mail]> wrote:
>> 
>> >The draft currently asks whether the Tube ID should be scoped to a single source address or not.  My preference is to stick to a single 5-tuple now, as in the common case, the path may change with a change in source address.
>> >
>> >There is still some chance that a changing privacy address in V6-land would not, but let's not optimize for that just now.
>> 
>> I'm ok with that for now.  The next question then is if the Tube ID adds any value at all?  Is the source sockaddr_in[6] enough?  We could save some bytes.
> ​I don't think it is enough, because you still need to disambiguate flows that go between the same endpoints but are different tubes.  The explicit identifiers allow you to make statements about their relationships more easily (think about the ordinal ranking use case, where a video app wants to say "this thumbnail flow is more important than the full-resolution flow")

Also this; it allows some spud endpoint users to group sub flows and bind properties in interesting ways.

(I was actually wondering if 64 bits are enough to allow enough subflow points while still keeping the active tube id space small)

> ​These aren't the bytes to save, in my opinion.
> 
> Ted​
> 
> 
>  
>> --
>> Joe Hildebrand
> 
> _______________________________________________
> Spud mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/spud