Re: [Anima-signaling] Unsolicited discovery [was GRASP issue 50: Rapid mode]

[Brian E Carpenter <[e-mail]>]

Hi Bing,
On 12/08/2016 19:11, Liubing (Leo) wrote:
> I folded the GRASP DT mailing list into the Anima folder. So I missed the large amount of recent discussion. I'll catch up.
> 
> Regarding Rapid Mode, when Sheng and I originally designed it, it was mostly inspired by the DHCP 2-message VS. 4-message. 
> Me too cannot blow the efficiency. But I think maybe the following scenarios make sense:
> 1. As Toerless commented on the mic: in IoT scenario, especially the on the lightweight and lossy links.

Yes, although on such links we may need CoAP in any case.

> 2. On some "heavy" nodes, which act as kind of central nodes, might reduce the number of maximum concurrent sessions.

Yes, that should be true for synchronization because it saves starting a thread and port for
the request/response. I'm not sure for negotiation, because that will normally be several
messages in each direction so we probably need a separate thread anyway.

I don't think there's anything wrong with the design of Rapid mode. The real question
is whether we want to have it as a MUST implement.

Regards
    Brian

> 
> Best regards,
> Bing
> 
>> -----Original Message-----
>> From: Anima-signaling [mailto:[e-mail]] On Behalf
>> Of Brian E Carpenter
>> Sent: Wednesday, August 03, 2016 4:44 AM
>> To: Toerless Eckert
>> Cc: Anima signaling DT
>> Subject: Re: [Anima-signaling] Unsolicited discovery [was GRASP issue 50:
>> Rapid mode]
>>
>> I'll be honest, I think Bing and I disagree about this. I see Rapid Mode as a
>> complication for quite a small gain in efficiency - there's actually more to be
>> gained by running GRASP over UDP instead of TCP, I think. So I probably
>> agree with you. But I think Bing sees more value in Rapid Mode than we do.
>>
>>     Brian
>>
>> On 02/08/2016 21:25, Toerless Eckert wrote:
>>> Thats fine too. Can we walk through an example set of elements in such
>>> a message ?
>>>
>>> Also: If we use Flood Synchronization, it sounds as if we would not
>>> need Rapid mode to make that work. In which case my other comment re.
>>> Rapid Mode might apply: unless we find a good example use-case within
>>> our existing or likely near term charter scope, maybe move Rapid mode
>> down into an appendix ?
>>>
>>> Cheers
>>>     Toerless
>>>
>>> On Tue, Aug 02, 2016 at 05:18:53PM +1200, Brian E Carpenter wrote:
>>>> On 01/08/2016 21:19, Toerless Eckert wrote:
>>>>
>>>> ...
>>>>>
>>>>> - Aka: I could happily let go of rapid-mode for now in the general case
>>>>>   (move into appendix as "candidate extension"), and just figure out
>> how
>>>>>   we can include unsolicited synchronization data into unsolicited
>> discovery
>>>>>   messages.
>>>>
>>>> We theoretically had unsolicited discovery responses in the design
>>>> for a while but never worked out details. They vanished when we added
>>>> Flood Synchronization which is an unsolicited announcement. It sends
>>>> an objective tagged with its source address. It's pretty much the same
>> thing you are asking for.
>>>>
>>>>     Brian
>>>
>>
>> _______________________________________________
>> Anima-signaling mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/anima-signaling
>