[address-policy-wg] Re: /48 or /56 to 'home' end-sites?

quite sure there was a discussion about this some months back and most 
people could agree on that a /56, maybe even /52 was a better size than 
/48. But maybe it was on ipv6-wg at ripe.net ?



On Wed, 7 Dec 2005, Jeroen Massar wrote:
> leo vegoda wrote:
> 
> [ Cut from the end, to reply on the former thread,
>   replyers might want to snip this part ]
> 
> > On 6 Dec 2005, at 11:55, Tim Chown wrote:
> >> Is this the
> >> way it's meant to be, or should the ISP owning the /32 only need to
> >> report usage when asking for more space itself?
> >
> > I think someone else mentioned that a sub-allocation would have worked
> > quite nicely in a case like this. It's worth noting that the current
> > policy does not require an LIR to get approval before making a
> > sub-allocation of any size.
> 
> It's also the way SixXS gets assigned a prefix, usually a /40, for a PoP
> from an ISP to be used for that ISP's PoP and manages this for that ISP,
> allocating toward endusers. One could see this as if one is 'outsourcing
> tunneled IPv6 end-user connectivity'.
> 
> 
> [ New Thread ]
> 
> > On 6 Dec 2005, at 11:55, Tim Chown wrote:
> <SNIP>
> >> We noted one knockon effect of RIPE policy.   I don't know the full
> >> details, but essentially for a tunnel broker service we wanted to offer
> >> a /48 to end sites out of an existing /32, but were unable to do so
> >> because the 'paperwork' to be sent on to RIPE-NCC for each /48 was needed
> >> in advance for the ISP owning the /32 to allocate a (say) /40 to the
> >> broker service, and that added a notable hurdle.   So we ended up using
> >> a /48 for the broker and allocating /56 and /64 blocks.
> 
> /56's are most likely enough for most 'home end-site'. A single /56
> provides 2^(64-56 = 8) = 256 /64's.
> 
> Let's introduce some terms:
> 
> end-site: a location where IP connectivity is delivered and managed by a
> single "IT/IS staff"/person.
> ISP: organization that provides connectivity to end-site.
> 
> Then we get two types of endsites:
> 
> 'home end-site': a location (residence) where living
> (fun,sleeping,eating) is the prime objective.
> 
> 'work end-site': a location where work is the prime objective and no
> sleeping  is done.
> 
> Of course there are a couple of nice things like hotels or resthouses,
> but those are delivering work to let the people live there, people there
> are being serviced to get living conditions (internet is a a requirement
> for that don't you think? :) etc. Another one could be home-workers.
> Maybe the city planning could determine it better, in many buildings you
> are not allowed to spend the night as they where not planned to be used
> for that.
> 
> 
> Currently both kinds of end-sites get /48's. For the work end-site the
> /48 should suffice all those sites as defined by the current policy.
> 
> For home end-sites I believe that a /48 really is way too much.
>  I have to see, let alone dream up, a home where even 25 separate
> networks would be used. Personally(*) I even ditched the routed network
> thing completely even though I effectively could setup at least 4 links,
> separately routed, thus 4x /64. Bridging seemed much easier in my case.
> I personally don't home users ever go over the 256 /64's, thus a /56
> should be sufficient. Of course future might change but still. Routing
> implies that people would configure it or that it would be autonomicly
> configured which is not available yet either. Prefix Delegation might
> work, but on 256 levels?
> 
> There where before a couple of other people noting the change of HD
> ratio, though I haven't seen much about that discussion recently, Tim's
> note above reminded me of it. In SixXS we typically use a /40 for a PoP,
> after 255 subnets (the first /48 is for tunnels) this assignment is
> full. In Tim's case with the same amount of users he will only use a
> single /48, which is 1/256th of what we just "burned". As noted above I
> personally believe that a /48 for 'work' sites is good. They can then
> always easily move between sites, and this will for certain be enough
> for 99.99% (or so) of the endsites that fall in this category. For a
> home-site situation a full /48 is IMHO really overkill, as noted above.
> 
> One of the reasons for saying 'all endsites a /48' was that all ISP's
> would give everybody a /48, renumbering would then not involve
> replanning onces network because one didn't get enough IP space.
> Creating a difference for home and work sites could only cause a problem
> when a work site becomes a home site, but I don't see that happening.
> Home to work, in case that happens, would get more address space at that
> point, thus that is not an issue either (except for the renumbering but
> let's not think about that, that is a different ballpark ;)
> 
> 
> Is it maybe time to look at this /48 policy and change it in the
> direction of the above that home endsites get a /56 instead of a full
> blown /48 which they will never use. Or do people think that it is fine
> and that we should not bother here at all?
> 
> Greets,
>  Jeroen
> 
> * = http://unfix.org/~jeroen/network/ in case one is wondering.
> 
> 

-- 


------------------------------
Roger Jorgensen              |
rogerj at stud.cs.uit.no        | - IPv6 is The Key!
http://www.jorgensen.no      | roger at jorgensen.no
-------------------------------------------------------