[MMUSIC] ICE-bis: pacing of ICE connectivity checks

[Ari Keränen <[e-mail]>]

Hi all,

Trying to close the open issue on the pacing of ICE connectivity checks, 
we still need to come up with the "new conservative default value" for 
the minimum pacing time with non-RTP traffic.

When this was discussed at the Orlando meeting, I proposed 100 ms (see 
[1] and [2]) but during the meeting there was a question that why could 
not we make this even smaller (e.g., 20 ms).

If we think about the worst case scenario with maximum length username 
fragment (256 bytes), that results in maximum size ICE connectivity 
check packets of about 330 bytes. Add a few extensions, rounding it up 
to 350 bytes, we're still at 35 ms "safe" pacing interval if we follow 
the rationale proposed by Cullen in [1] (i.e., similar bandwidth 
requirement as with G.711).

So, to be "conservative enough", how about 50 ms RECOMMENDED pacing 
value if there's no better knowledge of the network? And if there is, 
scale accordingly, but not below 20ms.


Cheers,
Ari

[1] http://www.ietf.org/mail-archive/web/mmusic/current/msg10448.html
[2] 
http://www.ietf.org/proceedings/86/minutes/minutes-86-mmusic#_Interactive_Connectivity_Establishment