On Thu, Mar 27, 2014 at 3:38 PM, Dave Thaler <[email protected]> wrote: >> -----Original Message----- >> From: Softwires [mailto:[email protected]] On Behalf Of Cb B >> Sent: Thursday, March 27, 2014 10:58 AM >> To: [email protected] >> Subject: [Softwires] draft-byrne-v6ops-clatip-01 >> >> Hi Softwires, >> >> Ales presented draft-byrne-v6ops-clatip-01 in softwires at the last IETF >> meeting. >> >> I am attempting to have this I-D adopted by v6ops, but v6ops requested >> feedback from softwires first. >> >> Pertaining to the minutes, i would like to address some topics to make sure >> it >> is ok for v6ops to move forward with adoption >> >> https://tools.ietf.org/wg/softwire/minutes?item=minutes-89-softwire.html >> >> The addresses, both in DS-lite and 464xlat, never appears on the wire so >> there is no chance of overlap or collision. > > Disagree, that conclusion doesn't follow (and in my experience it's wrong). > Overlap/collision happens when there are two interfaces on the same host > (even if they're not in use simultaneously). The collisions can affect > the routing table (if the host implements in such a way), ACLs like in > host firewall policies and such, and various application-layer uses. >
Ah, i see your point. If the host is itself both a B4 and a CLAT at the same time, then this collision may occur within the host, not on the wire. > It's fine to specify use as the default range (e.g. for 464xlat or DS-lite) > but > important to never constrain it to only that range, assuming the range is made > non-DS-lite specific. > > -Dave Is there such a constraint that would cause a problem? Looking at RFC6333 and draft-byrne-v6ops-clatip, i see that RFC6333 says the B4 SHOULD use 192.0.0.2. To a rational person, a good reason to not use 192.0.0.2 is that it is in use for a CLAT interface on the same host, which fits with the SHOULD wording. Is there some text that you could suggest that may clarify this situation in draft-byrne-v6ops-clatip or is it ok for v6ops to adopt as-is? As it stands, the I-D simply says that 192.0.0.0/29 will be generalized without making any further statements how addresses may be used within that range. CB _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
