Pekka and Gert,
Pekka Savola wrote: I don't think university-like organizations ever need more than a /48. One and the only really problematic thing is if they provide access to students/staff/etc., e.g. via DSL, dial-up, dorms or what. /64 would usually be ok (except very large universities and the like), /48 would not.
Universities are ISPs, and it is perfectly legitimate to assign /48s to students. Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48". Yes, they're going to use only two or three subnets possibly with only with one host each but they do need more than a /64 and I do not think that allocating anything between a /48 and a /64 is debatable at this point.
How do you think that e.g. Nokia, Cisco or Microsoft will do this? I think everyone agrees that /48 should be very well enough,
Pekka, I am sorry but this is wrong. One of the valuable participants we have in ipv6mh is Craig Huegen, the lead network architect for Cisco, and the _first_ question he asked about the protocols we are developing is how they would work for people that need more than a /48. With an HD of 0.8 Cisco needs between a /46 and a /45.
Gert Doering: So with some reason instead of a "this can't work!" attitude, I think this can work well - if one insists on doing non-useful things, it will break (but yes, this is a problem with the "one site" = /48 rule, because it's too vague to work unless people are reasonable).
Yes, the "one site" = /48 rule leads to terrible HD, but a university with 50K students would indeed fit within a /32. The question is: will people be reasonable. Michel.
On Sat, 11 May 2002, Michel Py wrote:
Pekka Savola wrote: I don't think university-like organizations ever need more than a /48. One and the only really problematic thing is if they provide access to students/staff/etc., e.g. via DSL, dial-up, dorms or what. /64 would usually be ok (except very large universities and the like), /48 would not.
Universities are ISPs
Usually not entitled to address space, based on current policy (not being LIR, peering reqs, ...). Universities may not even have an AS number.
, and it is perfectly legitimate to assign /48s to students.
Sure, if there is space to assign it from..
Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48".
Or use 6to4.
Yes, they're going to use only two or three subnets possibly with only with one host each but they do need more than a /64 and I do not think that allocating anything between a /48 and a /64 is debatable at this point.
Personally I advocate /48, but /64 is still good (multi-link subnet routing provides some missing features, and has no prefix assigning problems for John Doe).
How do you think that e.g. Nokia, Cisco or Microsoft will do this? I think everyone agrees that /48 should be very well enough,
Pekka, I am sorry but this is wrong. One of the valuable participants we have in ipv6mh is Craig Huegen, the lead network architect for Cisco, and the _first_ question he asked about the protocols we are developing is how they would work for people that need more than a /48. With an HD of 0.8 Cisco needs between a /46 and a /45.
I'd like to see how they're going to spend them. Note: if you use /64 for Point-to-Point links, this may be quite easy.
Gert Doering: So with some reason instead of a "this can't work!" attitude, I think this can work well - if one insists on doing non-useful things, it will break (but yes, this is a problem with the "one site" = /48 rule, because it's too vague to work unless people are reasonable).
Yes, the "one site" = /48 rule leads to terrible HD, but a university with 50K students would indeed fit within a /32. The question is: will people be reasonable.
As I read the table in global-ipv6-assign-2002-04-25.txt, /32 could provide addresses for 7132 users (assuming nothing is required for infrastructure). So at least /28 would be required. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
Hi, On Sat, May 11, 2002 at 08:28:57PM +0300, Pekka Savola wrote:
Yes, the "one site" = /48 rule leads to terrible HD, but a university with 50K students would indeed fit within a /32. The question is: will people be reasonable.
As I read the table in global-ipv6-assign-2002-04-25.txt, /32 could provide addresses for 7132 users (assuming nothing is required for infrastructure).
So at least /28 would be required.
As a side note: nothing prevents you from making better usage of a /32 - 7100 users is the minimum value you have to document to get a bigger block. Depending on the way the internal network hierarchy works (if it has few levels, and all entities at one level are similarily sized) it is perfectly allowed to utilize 20.000 /48s or more :-) - it all depends on your network structure. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 45114 (45077) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299
In your previous mail you wrote: Or use 6to4. => no, this doesn't solve anything because someone has to provide the 6to4 relay. IMHO 6to4 should never be recommended to an organization by the definition of what is an organization. What we want is one IPv6 Internet, not two! Regards Francis.Dupont@enst-bretagne.fr
On Sun, 12 May 2002, Francis Dupont wrote:
In your previous mail you wrote:
Or use 6to4.
=> no, this doesn't solve anything because someone has to provide the 6to4 relay. IMHO 6to4 should never be recommended to an organization by the definition of what is an organization. What we want is one IPv6 Internet, not two!
Did you miss the context: [Michel:]
Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48".
If *students* want to try this (having no practical experience about IPv6), *by far* the easiest way is 6to4. Nothing changes that. 6to4 relays are irrelevant from the addressing, and ease-of-use point of view. I wouldn't recommend 6to4 be used by organizations who want to really use IPv6 though. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
In your previous mail you wrote:
Or use 6to4.
=> no, this doesn't solve anything because someone has to provide the 6to4 relay. IMHO 6to4 should never be recommended to an organization by the definition of what is an organization. What we want is one IPv6 Internet, not two!
Did you miss the context: [Michel:]
Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48".
If *students* want to try this (having no practical experience about IPv6), *by far* the easiest way is 6to4. Nothing changes that. => if they are students as common customers with an IPv4 only ISP then 6to4 is a solution, in all other cases 6to4 should not be recommended. 6to4 relays are irrelevant from the addressing, and ease-of-use point of view. => 6to4 relays are the drawback of 6to4, in fact everybody should worry at the exception of IPv4 only ISPs (i.e. exactly the case where 6to4 should be recommended). I wouldn't recommend 6to4 be used by organizations who want to really use IPv6 though. => for an organization configured (so managed) tunnels are the proper solution until an ISP is proposing a native IPv6 connectivity. Regards Francis.Dupont@enst-bretagne.fr
On Sat, 11 May 2002, Pekka Savola wrote:
On Sat, 11 May 2002, Michel Py wrote:
Pekka Savola wrote: I don't think university-like organizations ever need more than a /48. One and the only really problematic thing is if they provide access to students/staff/etc., e.g. via DSL, dial-up, dorms or what. /64 would usually be ok (except very large universities and the like), /48 would not.
Universities are ISPs
Usually not entitled to address space, based on current policy (not being LIR, peering reqs, ...). Universities may not even have an AS number.
Interesting question is getting the number of them that *already* have one! Of course i know some cases... But discussing this at this level, gives a good idea about how NRENs (that have ASNs, and *are* ISPs) should solve their eventual "numbering" problem...
, and it is perfectly legitimate to assign /48s to students.
Sure, if there is space to assign it from..
NRENs should have it, assign them to universities and then universities can distribute it... "Direct" assign by NRENs to students doesnt seem very sensible...
Pekka, I am sorry but this is wrong. One of the valuable participants we have in ipv6mh is Craig Huegen, the lead network architect for Cisco, and the _first_ question he asked about the protocols we are developing is how they would work for people that need more than a /48. With an HD of 0.8 Cisco needs between a /46 and a /45.
I'd like to see how they're going to spend them.
Note: if you use /64 for Point-to-Point links, this may be quite easy.
We still dont have a clear guideline/BCP on this???
Gert Doering: So with some reason instead of a "this can't work!" attitude, I think this can work well - if one insists on doing non-useful things, it will break (but yes, this is a problem with the "one site" = /48 rule, because it's too vague to work unless people are reasonable).
Yes, the "one site" = /48 rule leads to terrible HD, but a university with 50K students would indeed fit within a /32. The question is: will people be reasonable.
On a normal day, i should say: NO! Some (misinformed) people tend to look at IP(whatever) addresses as a resource. Unfortunately this "some people" tends to be the majority... :-( ./Carlos "Networking is fun!" ------------------- http://www.fccn.pt <cfriacas@fccn.pt>, CMF8-RIPE, CF596-ARIN, Wide Area Network Workgroup F.C.C.N. - Fundacao para a Computacao Cientifica Nacional fax: +351 218472167
In your previous mail you wrote:
I don't think university-like organizations ever need more than a /48. One and the only really problematic thing is if they provide access to students/staff/etc., e.g. via DSL, dial-up, dorms or what. /64 would usually be ok (except very large universities and the like), /48 would not.
Universities are ISPs, and it is perfectly legitimate to assign /48s to students. Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48". Yes, they're going to use only two or three subnets possibly with only with one host each but they do need more than a /64 and I do not think that allocating anything between a /48 and a /64 is debatable at this point. => one detail which is never called to mind here is an university which gives internet access to its students *outside* the university (i.e. in a context where students can need a /48) should be sued by all commercial ISPs for illegal competition using public money... So universities can't be real ISPs by them selves, and to come back to the first topic, if you are an ISP or an ISP-like organization with a 2 year plan for 200 or more IPv6 /48 customers, the 3 BGP peers with the default-free routing table rule should not be a problem. The real problem is today an organization is supposed to be "large enough" to get a sub-TLA from its RIR or to be connected to such an organization which delegates a prefix. The case where the organization providing the connectivity doesn't provide a prefix too is both against the ideas of "IPv6 Address Allocation and Assignment Global Policy" and perhaps its letter. Regards Francis.Dupont@enst-bretagne.fr
On Sun, 12 May 2002, Francis Dupont wrote:
In your previous mail you wrote:
I don't think university-like organizations ever need more than a /48. One and the only really problematic thing is if they provide access to students/staff/etc., e.g. via DSL, dial-up, dorms or what. /64 would usually be ok (except very large universities and the like), /48 would not.
Universities are ISPs, and it is perfectly legitimate to assign /48s to students. Invariably, each class I teach IPv6 subnetting, students ask me "ok, now what if want to try this?" and my answer is "go to freenet6 and get a /48". Yes, they're going to use only two or three subnets possibly with only with one host each but they do need more than a /64 and I do not think that allocating anything between a /48 and a /64 is debatable at this point.
=> one detail which is never called to mind here is an university which gives internet access to its students *outside* the university (i.e. in a context where students can need a /48) should be sued by all commercial ISPs for illegal competition using public money...
The students can need a /48 for example to be able to easily connect to both commercial ISP and university (the same prefix length argument). Giving out /48's in universities is valid, but IMO /64's would also be quite enough. Basically we'd end up with one /48 for main use and 1-2 /48's for dial-up/DSL/etc.
So universities can't be real ISPs by them selves, and to come back to the first topic,
This is in practise a valid assumption, but I don't think the way you derived it is valid. That is: there is no law (usually) which says the university should give poorer service to its students, staff etc. than ISP's. I really can't see Universities being sued, except if they provide services to those that haven't enrolled or something.
if you are an ISP or an ISP-like organization with a 2 year plan for 200 or more IPv6 /48 customers, the 3 BGP peers with the default-free routing table rule should not be a problem.
You took a different approach here: if Big_Enough_ISP then 3 BGP peers and entries in DFZ should not be a problem This is new (perhaps this is the case, perhaps not. I can easily think of scenarios where small dial-up ISP's have 200 customers but no 3 BGP peers). But as was pointed out: if 3 BGP peers and entries in DFZ may not be Big_Enough_ISP from Global Policy point of view This what started this debate: NREN's. Note: in some small countries (Denmark?) there may not be that many Exchange Points or commercial operators that such a peering is possible. Difference is there.
The real problem is today an organization is supposed to be "large enough" to get a sub-TLA from its RIR or to be connected to such an organization which delegates a prefix. The case where the organization providing the connectivity doesn't provide a prefix too is both against the ideas of "IPv6 Address Allocation and Assignment Global Policy" and perhaps its letter.
That is a real problem. Usually probably caused by the fact that a) upstream organization does not care (about IPv6 at this point at least), or b) they can't really get a block anyway (or it is too difficult, and they don't want to go through all the troubles) due to too strict policies. In these cases, they probably end up with 6bone pTLA. Which is probably for the good for now. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
In your previous mail you wrote: The students can need a /48 for example to be able to easily connect to both commercial ISP and university (the same prefix length argument). => I don't believe in a /48 per student in an university framework, nor in commercial/academic dual homing.
So universities can't be real ISPs by them selves
This is in practise a valid assumption, but I don't think the way you derived it is valid. That is: there is no law (usually) which says the university should give poorer service to its students, staff etc. than ISP's. => there are a lot of laws about what may be done with public money and they give either an AUP which makes the network useless (a standard issue with "research" networks) or some kind of "collaboration" between a commercial ISP (in Europe the "historic" one) and the university. In the second case, there is no more a need for commercial/academic dual homing (a backdoor between the ISP and the internal network of the university is enough) and the "by them-selves" is removed. I really can't see Universities being sued, except if they provide services to those that haven't enrolled or something. => in this part of Europe I saw ISPs put pressure on universities and even local communities in order to "convince" them to not provide services. But even without this kind of practice, to provide ISP services is not the job of an university and should be avoided when there are other solutions. BTW I don't believe this applies to an USA large university/ campus.
if you are an ISP or an ISP-like organization with a 2 year plan for 200 or more IPv6 /48 customers, the 3 BGP peers with the default-free routing table rule should not be a problem.
You took a different approach here: if Big_Enough_ISP then 3 BGP peers and entries in DFZ should not be a problem This is new (perhaps this is the case, perhaps not. I can easily think of scenarios where small dial-up ISP's have 200 customers but no 3 BGP peers). => small == not big (:-)? IMHO here big enough is tear-N with N very small, and don't forget the 200 customers have to be IPv6 customers. But as was pointed out: if 3 BGP peers and entries in DFZ may not be Big_Enough_ISP from Global Policy point of view This what started this debate: NREN's. Note: in some small countries (Denmark?) there may not be that many Exchange Points or commercial operators that such a peering is possible. => I think the issue is not with UNI-C but with Nordunet which is a NREN and doesn't follow the standard ISP way to do business (of course, a NREN doesn't do business :-). Difference is there.
The real problem is today an organization is supposed to be "large enough" to get a sub-TLA from its RIR or to be connected to such an organization which delegates a prefix. The case where the organization providing the connectivity doesn't provide a prefix too is both against the ideas of "IPv6 Address Allocation and Assignment Global Policy" and perhaps its letter.
That is a real problem. Usually probably caused by the fact that a) upstream organization does not care (about IPv6 at this point at least), or b) they can't really get a block anyway (or it is too difficult, and they don't want to go through all the troubles) due to too strict policies. => IMHO the main reason of this problem comes from the fact that Nordunet is a NREN, not a real/standard ISP. In these cases, they probably end up with 6bone pTLA. Which is probably for the good for now. => I've got the same conclusion. Regards Francis.Dupont@enst-bretagne.fr
On Mon, 13 May 2002, Francis Dupont wrote:
In your previous mail you wrote:
The students can need a /48 for example to be able to easily connect to both commercial ISP and university (the same prefix length argument).
=> I don't believe in a /48 per student in an university framework, nor in commercial/academic dual homing.
I don't think it'll work all that well either, but some do; else "it's easy for NREN to reach 200 /48's" would never have come up.
This is in practise a valid assumption, but I don't think the way you derived it is valid. That is: there is no law (usually) which says the university should give poorer service to its students, staff etc. than ISP's.
=> there are a lot of laws about what may be done with public money and they give either an AUP which makes the network useless (a standard issue with "research" networks) or some kind of "collaboration" between a commercial ISP (in Europe the "historic" one) and the university. In the second case, there is no more a need for commercial/academic dual homing (a backdoor between the ISP and the internal network of the university is enough) and the "by them-selves" is removed.
How well this applies to everywhere in the RIPE region is questionable.
I really can't see Universities being sued, except if they provide services to those that haven't enrolled or something.
=> in this part of Europe I saw ISPs put pressure on universities and even local communities in order to "convince" them to not provide services. But even without this kind of practice, to provide ISP services is not the job of an university and should be avoided when there are other solutions. BTW I don't believe this applies to an USA large university/ campus.
I haven't heard about this at all here in Finland at least. So stating this kind of thing as a fact (applicable in all the RIPE areas) is a bit questionable.
if you are an ISP or an ISP-like organization with a 2 year plan for 200 or more IPv6 /48 customers, the 3 BGP peers with the default-free routing table rule should not be a problem.
You took a different approach here:
if Big_Enough_ISP then 3 BGP peers and entries in DFZ should not be a problem
This is new (perhaps this is the case, perhaps not. I can easily think of scenarios where small dial-up ISP's have 200 customers but no 3 BGP peers).
=> small == not big (:-)? IMHO here big enough is tear-N with N very small, and don't forget the 200 customers have to be IPv6 customers.
From the global policy point of view, any customer is fine. With a time span of 2 years, you can convince RIPE/ARIN/APNIC that the customers may get interested of IPv6.
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here. But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
But as was pointed out:
if 3 BGP peers and entries in DFZ may not be Big_Enough_ISP from Global Policy point of view
This what started this debate: NREN's. Note: in some small countries (Denmark?) there may not be that many Exchange Points or commercial operators that such a peering is possible.
=> I think the issue is not with UNI-C but with Nordunet which is a NREN and doesn't follow the standard ISP way to do business (of course, a NREN doesn't do business :-).
Peter Juul can comment better here if necessary, but the chain is: NORDUnet (transit for Nordic NREN's) Forskningsnettet (NREN for Denmark) UNI-C (University) The issues, as far as I understand it, are that: 1) NORDUnet is an "transit for NREN's" and does not provide any services or address space; NORDUnet does not have 200 customers. 2) Forskningsnettet (which is basically same as UNI-C I think) does not have have 3 peering partners?; Forskningsnettet does not have 200 customers (Universities etc.) either. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
On Mon, May 13, 2002 at 12:59:51PM +0300, Pekka Savola wrote:
2) Forskningsnettet (which is basically same as UNI-C I think)
Uni-C is a government-owned company that amongst other things takes care of the Forskningsnettet.
does not have have 3 peering partners?
We do and more. We just don't have three "ANY"-peering partners, which is what the DFZ demand boiled down to. That, however, doesn't really matter now that the new policy is in place. Peter B. Juul, Uni�C (PBJ255-RIPE)
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ? -- Xavier Henner Responsable de l'exp�rimentation IPv6 Nerim -- Fournisseur d'acc�s � Internet URL: <http://www.nerim.net/>
On Mon, 13 May 2002, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
Le Mon, May 13, 2002 at 03:43:18PM +0300, Pekka Savola a ecrit:
On Mon, 13 May 2002, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23.
and ? There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48. If an ISP has 1 million custommers, you give it enough adress space for static allocation (DSL is *not really* dial up) The /23 seems to be *big* but it's smaller than a /8 in IPv4 Today, in France, the 2 larger networks, in term of adress space must be Renater (the local NREN, with a lot of universities and their empty /16) and France Telecom, the historical telco, which is the biggest ISP too. With IPv6, the first will just have the same size as today. (And honestly, they could have a /20) Renater will have the size of a small ISP (I don't see how they can use more than a /32 : their "customers" can't have another transit provider and the campus networks for students are just tolerated) -- Xavier Henner Responsable de l'exp�rimentation IPv6 Nerim -- Fournisseur d'acc�s � Internet URL: <http://www.nerim.net/>
On Mon, 13 May 2002, Xavier Henner wrote:
Le Mon, May 13, 2002 at 03:43:18PM +0300, Pekka Savola a ecrit:
On Mon, 13 May 2002, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23.
and ? There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48.
Sure, if you go ask IANA for more (considerably more). I'd like to see the faces of RIR hostmasters when a few DSL/dialup etc. ISP's hand in address applications for their whole DSL/dialup service line..
If an ISP has 1 million custommers, you give it enough adress space for static allocation (DSL is *not really* dial up) The /23 seems to be *big* but it's smaller than a /8 in IPv4
Depends on how you count :-). IPv6 /23 is roughly equivalent to IPv4 /7 (48-23=25, 32-25=7) assuming every customer would get one IPv4 address. I don't really see any of these /8's currently being used being all that densely populated.. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
At 17:53 +0300 13/5/02, Pekka Savola wrote:
On Mon, 13 May 2002, Xavier Henner wrote:
Le Mon, May 13, 2002 at 03:43:18PM +0300, Pekka Savola a ecrit:
On Mon, 13 May 2002, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23.
and ? There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48.
Sure, if you go ask IANA for more (considerably more).
I'd like to see the faces of RIR hostmasters when a few DSL/dialup etc. ISP's hand in address applications for their whole DSL/dialup service line..
I suspect they wouldn't even blink and just hit the "approve" button. It is does not appear realistic to expect ISPs to "interview" each customer and ask them whether they intend to subnet or not. Everyone will probably end up with a /48: it reduces costs for the ISP. In this context who are the RIRs to argue about what is a site and what is not a site when handling an LIR request? Anyway, the argument has been that if you give out a /32 to each ISP, you need 2**29 ISPs (a lot) to exhaust prefix format 001 (not counting 6to4 and 6bone space which also inhabit that space). The problem does not seem to be conservation, it is scalability of the routing mesh. Joao
On Mon, 13 May 2002, Joao Luis Silva Damas wrote:
There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48.
Sure, if you go ask IANA for more (considerably more).
I'd like to see the faces of RIR hostmasters when a few DSL/dialup etc. ISP's hand in address applications for their whole DSL/dialup service line..
I suspect they wouldn't even blink and just hit the "approve" button. It is does not appear realistic to expect ISPs to "interview" each customer and ask them whether they intend to subnet or not.
I suspect they _would_ blink as the first thing they'd have to do is ask for more space: /23 is the same amount as is currently assigned to e.g. RIPE _altogether_. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
At 18:14 +0300 13/5/02, Pekka Savola wrote:
On Mon, 13 May 2002, Joao Luis Silva Damas wrote:
There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48.
Sure, if you go ask IANA for more (considerably more).
I'd like to see the faces of RIR hostmasters when a few DSL/dialup etc. ISP's hand in address applications for their whole DSL/dialup service line..
I suspect they wouldn't even blink and just hit the "approve" button. It is does not appear realistic to expect ISPs to "interview" each customer and ask them whether they intend to subnet or not.
I suspect they _would_ blink as the first thing they'd have to do is ask for more space: /23 is the same amount as is currently assigned to e.g. RIPE _altogether_.
That's OK, there is apparently plenty more where we got it the first time. Joao
-- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
Hi Joao, On Mon, 13 May 2002 17:08:17 +0200 "Joao Luis Silva Damas" <joao@ripe.net> wrote:
It is does not appear realistic to expect ISPs to "interview" each customer and ask them whether they intend to subnet or not.
so, you think it's not realistic to ask the customer if he intends to subnet or not ? ipv4 view: why not to give all your dialup customers in ipv4 per default a /28 delegation, and the customer decides to do NAT, all in one subnet, or more subnets ? no, i would give everyone /64 per default, unless the customer says, "no, i want to do subneting", then i will give him a /48 without discussion. thats my point of view.
Everyone will probably end up with a /48: it reduces costs for the ISP.
ipv4 view: some years ago,....: Everyone will probably end up with a /24 hm...i think the cost point isn't a good argument, sorry. does it cost so much to do subnetting ? you have one block with /64 delegations and one with /48's. so where is the problem ? take a /64 give it to the customers, if he says no, thake the other one. Thanks Robert -- Blechinger Robert Cybernet AG - Networking email: rblechinger@cybernet-ag.net Phone: +49 89 99315 - 116 Fax: +49 89 99315 - 199
Hi, On Mon, May 13, 2002 at 04:42:02PM +0200, Xavier Henner wrote:
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23.
and ? There is enough adress space for that. DSL lines will be used for home networks, and home networks are one of the reasons for the /48. If an ISP has 1 million custommers, you give it enough adress space for static allocation (DSL is *not really* dial up) The /23 seems to be *big* but it's smaller than a /8 in IPv4
I agree on this. I think it is a GOOD thing to have static IPs on DSL, and a /48 might be overkill for most households, it is valuable to those that do more interesting things with their network. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 45114 (45077) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299
Pekka Savola <pekkas@netcore.fi> writes:
On Mon, 13 May 2002, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
With HD ratio of 80%, the operator would require a /23.
Right, so allocate a /23 for it. Unless I miscalculated, 2000::/8 allows a good 4 billion (4 * 10^9) /48 assignments with a HD ratio of 0.8. So you can afford to assign a /23 for such a DSL operator without the danger of running out of address space. Robert
On Mon, May 13, 2002 at 04:11:20PM +0000, Robert Kiessling wrote:
Right, so allocate a /23 for it.
Unless I miscalculated, 2000::/8 allows a good 4 billion (4 * 10^9) /48 assignments with a HD ratio of 0.8.
So you can afford to assign a /23 for such a DSL operator without the danger of running out of address space.
Especially since even in a scenario, where the world population is 12 billion and there are twice that many connections out there using /48, 24000 ISPs of a million customers would cover the world. That still leaves us with a ridiculous number of unused /23s. Peter B. Juul, Uni�C (PBJ255-RIPE)
Hi, On Mon, May 13, 2002 at 01:45:52PM +0200, Xavier Henner wrote:
If you *do* have 200 IPv6 customers *today* (those that pay for the service, not necessarily IPv6 service, but something), I'd surely consider one a real player here.
But an issue is about DSL operators with 1,000,000 IPv4 customers beginning to introduce IPv6 services..
I don't see any issue with DSL operators. Can you explain it ?
Well - 1 million customers with a /48 each makes a BIG allocation :-) (Of course I'd like to see a DSL service, especially to home users, with real and static IPv6 addresses) Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 45114 (45077) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299
On Mon, May 13, 2002 at 11:45:45AM +0200, Francis Dupont wrote:
=> I think the issue is not with UNI-C but with Nordunet which is a NREN and doesn't follow the standard ISP way to do business (of course, a NREN doesn't do business :-).
Let me just correct a slight misunderstanding here: Nordunet is a cooperation between a number of NRENs in the nordic countries such as Sunet (SE), Uninett (NO), FUNET (FI) and FSKNET (DK). FSKNET is run by Uni-C. Thus it is FSKNET that is the NREN and ISP. It's not very important, but we might as well get it right :-)
=> IMHO the main reason of this problem comes from the fact that Nordunet is a NREN, not a real/standard ISP.
I honestly have a hard time seeing the difference between an NREN that provides about 100K (rough estimate) users in about 200 institutions with IP, DNS and other network services, maintains its own backbone, peers at the relevant peering points and so on and a "regular ISP". The main difference is that there's a defined limit to our customer base, as we can't go and compete in the commercial market, but has to stay within the research and education world. That limit is hardly relevant to our place in the Internet fabric. Peter B. Juul, Uni�C (PBJ255-RIPE)
On Sun, 12 May 2002, Francis Dupont wrote:
=> one detail which is never called to mind here is an university which gives internet access to its students *outside* the university (i.e. in a context where students can need a /48) should be sued by all commercial ISPs for illegal competition using public money...
I think most NREN AUP's make this sort of activity unlikely; certainly in the UK you can't generally run commercial services over JANET, and you certainly can't resell academic bandwidth to external users.
So universities can't be real ISPs by them selves, and to come back to the first topic, if you are an ISP or an ISP-like organization with a 2 year plan for 200 or more IPv6 /48 customers, the 3 BGP peers with the default-free routing table rule should not be a problem.
I would hope that universities do not want or need a SubTLA?
The real problem is today an organization is supposed to be "large enough" to get a sub-TLA from its RIR or to be connected to such an organization which delegates a prefix. The case where the organization providing the connectivity doesn't provide a prefix too is both against the ideas of "IPv6 Address Allocation and Assignment Global Policy" and perhaps its letter.
I would envisage a university getting one /48 for its campus and another out of which to allocate staff/student dialup/etc - as Pekka says for this kind of activity (the university is not a full-blown ISP) the end user gets a /64. An interesting question is whether the types of devices that become deployed in campus departments keep the routing hierarchy similar for IPv6 as IPv4, or whether we'll see devices that offer a /64 per office, for example. Tim
In your previous mail you wrote:
=> one detail which is never called to mind here is an university which gives internet access to its students *outside* the university (i.e. in a context where students can need a /48) should be sued by all commercial ISPs for illegal competition using public money...
I think most NREN AUP's make this sort of activity unlikely => with a strict or really enforced AUP the network can become near useless (at least for students who have extra-academic interests :-). So usually the AUP is loose or not really enforced *and* a commercial ISP is implied into the MAN/WAN infrastructure and management. IMHO this is the least bad solution because the ISP has technical and man-power capacities which are not usual in an university and never in its missions. On the other side the ISP gets "subsidized" customers... certainly in the UK you can't generally run commercial services over JANET, and you certainly can't resell academic bandwidth to external users. => I shan't ask details about "generally" and "external" (:-). An interesting question is whether the types of devices that become deployed in campus departments keep the routing hierarchy similar for IPv6 as IPv4, or whether we'll see devices that offer a /64 per office, for example. => in the campus or the university LANs are better (easier, cheaper, etc) so I don't believe the issue exists. Regards Francis.Dupont@enst-bretagne.fr
Please excuse this lurker joining the conversation. I have a very little IPv6 experience so please give what I say the appropriate level of reliability.
From: Tim Chown <tjc@ecs.soton.ac.uk>
On Sun, 12 May 2002, Francis Dupont wrote:
=> one detail which is never called to mind here is an university which gives internet access to its students *outside* the university (i.e. in a context where students can need a /48) should be sued by all commercial ISPs for illegal competition using public money...
I think most NREN AUP's make this sort of activity unlikely; certainly in the UK you can't generally run commercial services over JANET, and you certainly can't resell academic bandwidth to external users.
Whilst this is one way to view the JANET Connection Rolicy and AUP (<http://www.ja.net/documents/connection_policy.pdf>, <http://www.ja.net/documents/use.html>) in practice there seems to be quite a bit more latitude. The getouts are in clauses that talk about things like "benefits to the Higher and Further Education and Research Council community". One possible application is in providing services to alumni rather than current students. I don't know if any precedents have been set on JANET yet but it's an obvious area.
So universities can't be real ISPs by them selves, ... [snip]
I would envisage a university getting one /48 for its campus and another out of which to allocate staff/student dialup/etc - as Pekka says for this kind of activity (the university is not a full-blown ISP) the end user gets a /64.
On JANET at least - I don't know about other NRENs - a University is allowed to offer onward connections (Sponsored connections), acting as a mini-ISP or proxy-PoP. There are benefits for the sponsored site to use address space which is separate from the sponsoring (University) site's, and in IPv4 on JANET this is generally enforced. Should this mean a third /48 for any such third party sites attached the University? Sam Wilson Network Services Division Computing Services, The University of Edinburgh Edinburgh, Scotland, UK
On 13 May 2002 Sam.Wilson@ed.ac.uk wrote:
Whilst this is one way to view the JANET Connection Rolicy and AUP (<http://www.ja.net/documents/connection_policy.pdf>, <http://www.ja.net/documents/use.html>) in practice there seems to be quite a bit more latitude. The getouts are in clauses that talk about things like "benefits to the Higher and Further Education and Research Council community". One possible application is in providing services to alumni rather than current students. I don't know if any precedents have been set on JANET yet but it's an obvious area.
I think this is now more than a shade off-topic (:-) but yes a university may feel it wants to offer connectivity to past students/staff. However the bulk of these will not live locally, nor will the university offer local rate dial-up nationally. In the UK case, UKERNA used to have an arrangement with U-NET that gave (current) students local rate dialup to a national service, and that was a recommended alternative to universities (or more likely the small colleges) who didn't want to manage their own dialup. But that was terminated (http://www.ja.net/services/jnds/). In terms of the get-outs, I expect that commercial spin-offs or products from the universities (e.g. a university selling intelligent widgets) may be hosted from a university with such a spin-off. However, at my own university we do this sort of activity via a commercial leased line, but I agree the rules have become more lax (we are in effect paying twice for bandwidth to be "clean") and that's why I used words like "generally" to Francis... the really fuzzy area is universities who use their academic connectivity to give students access from halls of residence, which if done with open ADSL-like rules may well let students run commercial services. Tim
participants (12)
-
Carlos Friacas -
Francis Dupont -
Gert Doering -
Joao Luis Silva Damas -
Michel Py -
Pekka Savola -
Peter B . Juul -
Robert Blechinger -
Robert Kiessling -
Sam.Wilson＠ed.ac.uk -
Tim Chown -
Xavier Henner