Re: Comments on draft-sijeon-dmm-use-cases-api-source-04

Le 20/07/2016 à 23:17, Seil Jeon a écrit :
> Hi Alex,
>
> Thanks for your review. Please find "SJ>" inline.
>
> Regards,
> Seil Jeon
>
> -----Original Message-----
> From: dmm [mailto:[email protected]] On Behalf Of Alexandre Petrescu
> Sent: Wednesday, July 20, 2016 6:42 PM
> To: [email protected]
> Subject: [DMM] Comments on draft-sijeon-dmm-use-cases-api-source-04
>
> Hi DMMers,
>
> I read draft-sijeon-dmm-use-cases-api-source-04 and I have a few comments:
>
>> Applications such as a text-based web browsing or information-centric
>> service, e.g. weather and stock information, as well as legacy
>> applications may belong to this category.  As many applications
>> require short-lived Internet connection without session continuity and
>> IP address reachability,
>
> These are _interactive_ web browsing applications - both the text-based such
> as lynx, and graphics-based such as firefox browsers, are interactive and
> needs a short-lived Internet connection, such as a short-lived TCP socket
> (user clicks, socket opens, retrieve page through that socket, socket
> closes, then user clicks again - that's interactivity).
>
> It is little qualified to call them 'text-based'.
>
> The same with 'information-centric' - there should be a better term for
> that.  It is typically data that is 'consumed' by the human user (as opposed
> to POSTing data, and as opposed to consumption by machines).  It would be
> better to talk about such consuming, rather than 'information-centric' which
> may make think of ICN: information-centric networking.
>
> SJ> O.K. I get your point. But some "interactive" applications need session
> continuity,

I doubt so?  Apps that need session continuity, IMHO, are typically 
non-interactive - the user may watch the video, listen the MP3 or the 
incoming audio part of the video call.  These would be little known as 
"interactive apps".

> so the term may not be in the correct meaning we'd like to
> express. How do you think "non-persistent IP connection" instead of them?

I think that is a good term.  Although it is little used outside DMM 
discussions.

We need somehow to relate more to terms like web browsers, streaming, 
which are widely understood outside DMM too.

>
>
>> The suggested flag, IPV6_REQUIRE_NON-PERSISTENT_IP, defined in
>> [I-D.ietf-dmm-ondemand-mobility] is used for expressing its preference
>> to the IP stack.
>
> This is a parameter to which function?  To socket ()?  ioctl ()?  Is there a
> brief example of 5 lines of C code about how to call it?
>
> SJ> It's a Socket API defined in the on-demand mobility draft. It explains
> what will be the normal operation in the given use case. Description about
> the on-demand Socket API is copied from
> https://tools.ietf.org/html/draft-ietf-dmm-ondemand-mobility-07

Is it a parameter to function socket ()?  ioctl ()?  I am just 
wondering.  These functions are standardized, but there may be others I 
forgot...

Alex

> ---------------------------------------------------------------------
> The following new flags are defined in this document and they shall
>    be used with Socket API in compliance with the [RFC5014]:
>
>    IPV6_REQUIRE_FIXED_IP /* Require a Fixed IP address as source */
>
>    IPV6_REQUIRE_SESSION_LASTING_IP /* Require a Session-lasting IP
>    address as source */
>
>    IPV6_REQUIRE_NON-PERSISTENT_IP /* Require a Non-persistent IP address
>    as source */
> ---------------------------------------------------------------------
>
>
>
>
>
>
> Alex
>
> _______________________________________________
> dmm mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/dmm
>
>