6to4 status (again)
Kevin Day
kevin at your.org
Fri Mar 1 18:56:32 CET 2013
On Mar 1, 2013, at 3:24 AM, Lorenzo Colitti <lorenzo at google.com> wrote:
> On Fri, Mar 1, 2013 at 12:39 AM, David Farmer <farmer at umn.edu> wrote:
> So PLEASE just don't go turning off 6to4 relays willy-nilly, turning them off when there is no more traffic is fine. But, just turning them off when there is traffic adds to bad user experience we are trying to correct.
>
> If you need to manage the traffic to your relay then change your BGP advertizements to make it look less attractive, or change the scope of your advertizements to cover only your direct customers, but don't just turn it off.
>
> Every time someone turns off a relay or limit the scope of the BGP advertisements they are reduce the total capacity of the system. The traffic doesn't disappear, it goes somewhere else, and with every relay it's more likely to be dropped. There's just no way to win this game.
That's been my concern with shutting our relay down.
http://bgp.he.net/net/192.88.99.0/24 shows 23 ASNs announcing 192.88.99.0/24. I'm sure he.net's list isn't complete, but I'm also guessing they're seeing some announcements that are "peers only" so not all of those are public. For the sake of argument, let's say there are 23 public relays in the world.
We're rapidly approaching 1gbps on ours. If I shut down, 1000mbps/23 = 43mbps will shift on average to all the others if it's a total even split. This is probably not enough to break any individual relay, but I know a lot of these relays are on commodity hardware that can't possibly be using 10GE gear. Bringing everyone ~5% closer to saturating their link can only happen so many times before everyone else throws in the towel as well.
If these are legitimate users that are using 6to4 due to old hardware or require 6to4 for reasons I'm unaware of, I'm prepared to set up additional resources to help out - maybe even getting more relays set up. If this usage really is only from torrent users who are using 6to4 to evade ISP detection and people using relays as DDoS amplifiers, I'm a lot more inclined to let it just die.
We also should maybe consider a more determined effort to get people off 6to4 - I'm imagining something akin to the "Please upgrade away from IE6" banners that some sites were using, if the client IP was in the 2002::/16 range. I have no idea how many people are using 6to4 and just don't even realize it, which is perpetuating the 'ipv6 is slow' mantra.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cluenet.de/pipermail/ipv6-ops/attachments/20130301/6e65fd06/attachment.htm>
More information about the ipv6-ops
mailing list