Dave - I'm responding to your e-mail but my post is aimed at the homenet WG as 
a whole, because your e-mail gives me an opportunity to pull on a couple of 
threads from my perspective as a DT member.

You suggested installing the software.  OK, I could install it on my home 
network, but I would gain exactly zero insight into the problem from that 
installation.  My home network has one router, and everything inside my home is 
bridged.

I do have native IPv6 (thanks, JohnB and Comcast) in my home, through a Linksys 
router with an old, internal software build from Cisco.  That router waited 
patiently until it got a DHCPv6-PD delegation from Comcast and then silently 
started advertising the prefix.  I didn't notice that the Comcast IPv6 service 
had been added to the tunnel service I was using until I happened to notice a 
new prefix on one of the devices on the home network.  I shut down the tunnel 
and everything just worked...a demonstration of graceful renumbering 
(admittedly not a very complicated demo, but one in a real, operational 
network) on running code.

So, as you suggest, I could go buy a couple of routers and run the software on 
a more complex topology.  How do I decide where to put those routers so I can 
learn anything about the protocols?  I don't ever intend to use any of my 
wireless as a transit link.  I have a wired link up to the second floor with a 
switch and an AP to improve coverage on the second floor, but no redundancy so 
there's no advantage to adding a router in front of the switch.  Where's the 
advantage in running a collection of self-configuring routers in my current 
network?

OK, I could invent a new topology that might take some advantage of the 
routers.  How do I choose that topology?  What are the requirements I'm trying 
to test?  Where do I find a couple of weeks to perform experiments?

Therefore, in my estimation, I *might* learn something more than I know now by 
spending a significant amount of my time or my employer's time in a series of 
experiments using deployments of the alternative routing protocols on 
hypothetical network topologies and use cases.  I don't see how that knowledge 
is going to help me solidify the requirements or answer the questions Russ 
talked about in his report from the DT in Prague.

- Ralph

> On Aug 5, 2015, at 8:48 AM 8/5/15, Dave Taht <[email protected]> wrote:
> 
> On Wed, Aug 5, 2015 at 2:34 PM, Ray Bellis <[email protected]> wrote:
>> 
>> 
>> On 05/08/2015 12:44, Dave Taht wrote:
>> 
>>> I would like to require the design team
>>> 
>>> *to actually install the software*.
>> 
>> Dave,
>> 
>> We've heard you before, but with the best will in the world we cannot
>> *require* IETF volunteers to put their own time and money into this.
> 
> Stay home and save the plane fare and hotel and food expenses.
> Attending a meeting costs 3k or so, plus the lost business from trying
> to stay awake in futile meetings.
> 
> That's what I've been doing. Pays for a lot of routers, that does.
> Gives me time to work on the remaining real problems, also, which
> include naming the ipv6 hex vomit more sanely and a multitude of other
> issues this wg has dropped the ball on.
> 
> If whoever is funding your travel is unwilling to let you expense a
> couple 100 dollar (tops) routers, in addition to allowing you to waste
> time attending meetings, I think they are investing in the wrong
> place.
> 
> Nobody funds my travel, and I guess, in part, that is why I am aware
> of the overwhelming difference in cost between actually developing and
> testing the software, vs attending meetings or putting up with the
> endless bikeshed emails.
> 
>> Heck, I can't even run the Homenet stack myself as my ISP doesn't yet
>> have native IPv6 and also wants extra money to get their edge router
>> running in bridge mode.
> 
> Nearly every form of ipv6 tunneling is supported by the openwrt
> software for your edge gateway.
> 
> dslite, 6in4, aiccu, 6to4, 6rd, and gre
> 
> are supported and have been supported *for years*. Numerous other
> methods of applying a ipv6 vpn over ipv4 also exist.
> 
> If you provider will not allow any of those through their crappy cpe,
> then perhaps this wg needs to consider that as a problem for this wg?
> 
> Sure, that first router requires some hand configuration for the
> needed tunnels. Additional ones do not, and can thus be tested with
> hnetd and your routing protocol of choice.
> 
> http://wiki.openwrt.org/doc/uci/network for more details as to setup.
> 
>> 
>> [and no, before you suggest it, changing ISPs is *not* an option].
> 
> Not being able to get native ipv6 to your location is no excuse.
> 
> 
>> Ray
>> 
> 
> 
> 
> -- 
> Dave Täht
> worldwide bufferbloat report:
> http://www.dslreports.com/speedtest/results/bufferbloat
> And:
> What will it take to vastly improve wifi for everyone?
> https://plus.google.com/u/0/explore/makewififast
> 
> _______________________________________________
> homenet mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/homenet

_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to