Global Crossing says it has deployed native IPv6. Also, TeliaSonera has picked Lucent to help it prepare for IPv6 service. http://www.techweb.com/wire/172300284 --Steven M. Bellovin, http://www.cs.columbia.edu/~smb
On Wed, Oct 12, 2005 at 10:33:42AM -0400, Steven M. Bellovin wrote:
Global Crossing says it has deployed native IPv6. Also, TeliaSonera has picked Lucent to help it prepare for IPv6 service.
The full GC PR is at; http://www.globalcrossing.com/xml/news/2005/october/10.xml (Full Disclosure; I'm an SNE with HEAnet). -- Colm MacCárthaigh Public Key: colm+pgp@stdlib.net
On Wed, 12 Oct 2005, Colm MacCarthaigh wrote:
On Wed, Oct 12, 2005 at 10:33:42AM -0400, Steven M. Bellovin wrote:
Global Crossing says it has deployed native IPv6. Also, TeliaSonera has picked Lucent to help it prepare for IPv6 service.
The full GC PR is at;
Umm.. "IPv6 [...] delivered over our global, MPLS-based backbone." It's not clear whether they're doing 6PE over their v4/MPLS backbone, running v6 in parallel to v4/MPLS or running v6/MPLS (I don't think vendors support this). At least one of these doesn't (IMHO) qualify as "native IPv6 [backbone]". -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
On Wed, Oct 12, 2005 at 10:00:23PM +0300, Pekka Savola wrote:
On Wed, 12 Oct 2005, Colm MacCarthaigh wrote:
On Wed, Oct 12, 2005 at 10:33:42AM -0400, Steven M. Bellovin wrote:
Global Crossing says it has deployed native IPv6. Also, TeliaSonera has picked Lucent to help it prepare for IPv6 service.
The full GC PR is at;
Umm..
"IPv6 [...] delivered over our global, MPLS-based backbone."
It's not clear whether they're doing 6PE over their v4/MPLS backbone, running v6 in parallel to v4/MPLS or running v6/MPLS (I don't think vendors support this).
At least one of these doesn't (IMHO) qualify as "native IPv6 [backbone]".
They are delivering native v6 sessions (both customer handoff and backbone links) via their Juniper core. I don't know what they're doing with the GSRs, or even how many of them they have left, but I di know that all of the Juniper-based v6 is native. The "MPLS-based backbone" stuff is just standard marketing fluff. -- Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)
Hi all, Take the opportunity to make a non commercial add ;-) Every day there are more and more news related to IPv6. I compile all them at http://www.ipv6tf.org. I also emails every Monday a summary, not sure if it will be good to send it also to this list ? Alternatively, you can register at the site and will get it, together with access to other sections. Regards, Jordi
De: "Steven M. Bellovin" <smb@cs.columbia.edu> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 10:33:42 -0400 Para: <nanog@nanog.org> Asunto: IPv6 news
Global Crossing says it has deployed native IPv6. Also, TeliaSonera has picked Lucent to help it prepare for IPv6 service.
http://www.techweb.com/wire/172300284
--Steven M. Bellovin, http://www.cs.columbia.edu/~smb
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
good news. but .... if you look at the recent ipv4 burn rate of ripe and apnic especially, we run out of v4 space in about three years. this should not be surprising, as it matches what frank was saying a decade ago at ale. so having dual stack backbones is very important. but ... four years from now, when marissa can't get v4 space from an rir/lir and so gets v6 space, she will not be able to use 99% of the internet because no significant number of v4 end hosts will have bothered to be v6 enabled because there was no perceived market for it. there will likely be a dangerous period between v4 exhaustion and significant v6 presence where v6-only folk will be in a very bad place. geoff's predictions for a very lively market in v4 space will seriously come into play. randy
On 10/12/05 3:13 PM, "Randy Bush" <randy@psg.com> wrote:
geoff's predictions for a very lively market in v4 space will seriously come into play.
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG? This could extend the lifetime of the IPv4 space significantly by promoting efficient use through economic incentives, provide positive economic incentives to move to v6 when needed, and eliminate the grey market. Proper controls could be put into place to prevent de-aggregation through utilization of the RIRs as clearing houses.
randy
On Wed, 12 Oct 2005, Daniel Golding wrote:
On 10/12/05 3:13 PM, "Randy Bush" <randy@psg.com> wrote:
geoff's predictions for a very lively market in v4 space will seriously come into play.
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG?
I, for one, would be very interesting in such a system. Distribution of commodities is almost universally done best by capital markets. Unfortunately I won't be at the next NANOG. -- Brandon Ross AIM: BrandonNRoss Director, Network Engineering ICQ: 2269442 Internap Skype: brandonross Yahoo: BrandonNRoss
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG?
I, for one, would be very interesting in such a system. Distribution of
commodities is almost universally done best by capital markets.
There is a slight problem here. Commodities are things which are bought and sold. In other words, one party has legal title to the commodity and transfers that legal title to another party. Since nobody has legal title to any IPv4 addresses, nobody can sell them in the first place. Of course you can get around this by selling the networks that use the IPv4 addresses, but then you are getting away from the realm of commodities. A commodity is a fairly generic product and networks are far from generic. --Michael Dillon
On Thu, 13 Oct 2005 Michael.Dillon@btradianz.com wrote:
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG?
I, for one, would be very interesting in such a system. Distribution of
commodities is almost universally done best by capital markets.
There is a slight problem here. Commodities are things which are bought and sold. In other words, one party has legal title to the commodity and transfers that legal title to another party. Since nobody has legal title to any IPv4 addresses, nobody can sell them in the first place.
That's exactly the change I've been advocating for years. Instead of continuing with this socialistic concept that IP space is somehow owned by everyone, we should, instead, give title for IP space and allow those titles to be bought and sold freely. Classic economics teaches of the tragedy of the commons. I can't think of too many things that look more like a commons than the current IP space. By my own best estimates, 50% of the allocated space today is wasted in one way or another, either it is used inefficiently by staticly addressing things that don't need to be static, hoarded to prevent organizations from having to make additional requests to an RIR, or legacy assignments where the orgs that have them have no incentive to give them up. Almost all of the exhaustion problems that are on the horizon are being directly caused by inefficient use of this scarce resource, certainly all of the above is solved by a capital market.
Of course you can get around this by selling the networks that use the IPv4 addresses, but then you are getting away from the realm of commodities. A commodity is a fairly generic product and networks are far from generic.
Again, converting to a capitalistic system is how we can stop this underhanded practice. -- Brandon Ross AIM: BrandonNRoss Director, Network Engineering ICQ: 2269442 Internap Skype: brandonross Yahoo: BrandonNRoss
Brandon Ross wrote:
On Thu, 13 Oct 2005 Michael.Dillon@btradianz.com wrote:
[IPv4 commodity trading]
That's exactly the change I've been advocating for years. Instead of continuing with this socialistic concept that IP space is somehow owned by everyone, we should, instead, give title for IP space and allow those titles to be bought and sold freely. Classic economics teaches of the tragedy of the commons. I can't think of too many things that look more like a commons than the current IP space.
IRL, I work for a very large company that controls a significant fraction of legacy Class B networks. Most of this control is due to multiple acquisitions and mergers. I would posit that the whole place could be run out of one /16, and the rest of it turned back in, or sold off, so that it could be reused, and delay the exhaustion.
By my own best estimates, 50% of the allocated space today is wasted in one way or another, either it is used inefficiently by staticly addressing things that don't need to be static, hoarded to prevent organizations from having to make additional requests to an RIR, or legacy assignments where the orgs that have them have no incentive to give them up.
Well, I doubt very much that I'd ever agree that static addresses are inefficient, but legacy assignments need to be readdressed. Until there is economic advantage in surrendering space, or disadvantage in keeping it, this will not change. I suspect that most of the organizations that have large legacy spaces give no more than passing thought to such things. If they thought that they were holding capital assets, on the other hand, they'd be in the market for selling within weeks. I've tried on my own to persuade folk into the surrender of IP space, and the immediate response is "Why should we?" That was two acquisitions ago; the answer (from them) hasn't changed. To my knowledge, it is the equivalent space of 35 legacy Class B networks. You do the math. -- There are two ways, my friend, that you can be rich in life. One is to make a lot of money and the other is to have few needs. William Sloane Coffin, "Letters to a Young Doubter"
On Wed, Oct 12, 2005 at 03:20:31PM -0400, Daniel Golding wrote:
On 10/12/05 3:13 PM, "Randy Bush" <randy@psg.com> wrote:
geoff's predictions for a very lively market in v4 space will seriously come into play.
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG?
This could extend the lifetime of the IPv4 space significantly by promoting efficient use through economic incentives, provide positive economic incentives to move to v6 when needed, and eliminate the grey market.
Proper controls could be put into place to prevent de-aggregation through utilization of the RIRs as clearing houses.
First of all, I'm still waiting to be convinced that there is actually an IP shortage at all. From the latest routing table analysis dump to nanog: Percentage of available address space announced: 38.6 Percentage of allocated address space announced: 58.1 Percentage of available address space allocated: 66.4
From where I sit, the perceived shortage is due to non-existant reclamation of unused resources, and financial incentives to create an artificial shortage. As much as I like to see capitalism solve problems, I don't think that opening up a market in selling legacy allocations is going to make things better.
It is one thing to have a legacy allocation sitting around "just incase", when the only value is reduced annoyance if you ever need to get more IP space in the future. It is another thing to have the allocation actually be worth something monitarily, and potentially worth a big something if you can manage to hold onto it until there is a REAL shortage (maybe even one that a legacy allocation owner can help create if they have any policy control, wink wink nudge nudge). Capitalism can only sort things out when there is a truely open market, which I don't think describes this situation at all. All I see is that in 3-4 years we will actually have to engage our collective brains again and start getting new IP allocations from a different source. It's not an exhaustion of IPv4 at all, it is just a next step in the evolution of the Internet. Call it recycling if you will. Investing a little bit of time and effort into figuring out the reclamation process now would save us a lot of grief a few years down the road. Why don't we start by going after the low hanging fruit, and pressure some non-corporate entities like the US government to return some of its legacy unused /8 allocations. I'm certain that someone with some historical BGP data could put together an analysis of who has not used their IP allocations at ALL within the last few years, still more low hanging fruit which we can take care of now. Of course, the last time I mentioned an unused /8 which should have been returned years ago on this list, the party in question started announcing it in BGP the next day. -- Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)
ras, all On Wed, Oct 12, 2005 at 04:38:53PM -0400, Richard A Steenbergen wrote:
of its legacy unused /8 allocations. I'm certain that someone with some historical BGP data could put together an analysis of who has not used their IP allocations at ALL within the last few years, still more low hanging fruit which we can take care of now. Of course, the last time I mentioned an unused /8 which should have been returned years ago on this list, the party in question started announcing it in BGP the next day.
the problem here is this: there is no guarantee that prefixes that are never seen in global tables are not used and deployed. for example, the US DoD has quite a lot of address space (pre-rfc-1918) deployed onto the SIPRNet, i believe. this is not routed to the public internet, but is in use. an argument could be made that one could ignore that space, since it is never intended to route publicly, but intentions change and address/prefix conflicts are bad. by saying this i don't intend to disagree with the general premise: there are tons of genuinely unused prefixes out there. the point is just that i doubt that there is an automated way to determine exactly which ones they are. -- _____________________________________________________________________ todd underwood director of operations & security renesys - interdomain intelligence todd@renesys.com www.renesys.com
by saying this i don't intend to disagree with the general premise: there are tons of genuinely unused prefixes out there. the point is just that i doubt that there is an automated way to determine exactly which ones they are.
depends on what you mean by automated. geoff's point, among other things, is that we will see social automation in action when the rirs/lirs can no longer allocate from an exhausted iana v4 pool. also to be noted is that rir statistics on who has what space are not in the best of shape, ripe's being particularly obfuscated. randy
On Wed, Oct 12, 2005 at 11:13:12AM -1000, Randy Bush wrote:
also to be noted is that rir statistics on who has what space are not in the best of shape, ripe's being particularly obfuscated.
*raising an eyebrow* Would you care to elaborate on that? Best regards, Daniel -- CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
On Wed, 12 Oct 2005 23:16:03 +0200, Daniel Roesen said:
On Wed, Oct 12, 2005 at 11:13:12AM -1000, Randy Bush wrote:
also to be noted is that rir statistics on who has what space are not in the best of shape, ripe's being particularly obfuscated.
*raising an eyebrow*
Would you care to elaborate on that?
Just guessing, but I think Randy is saying that not everybody is totally up-to-date on making sure all the SWIP data is correct....
* Daniel Roesen:
On Wed, Oct 12, 2005 at 11:13:12AM -1000, Randy Bush wrote:
also to be noted is that rir statistics on who has what space are not in the best of shape, ripe's being particularly obfuscated.
*raising an eyebrow*
Would you care to elaborate on that?
AFAIK, the status of EARLY-REGISTRATION space is still somewhat murky (my favorite topic 8-).
Percentage of available address space announced: 38.6
You misunderstand what IP addresse are. They have nothing whatsoever to do with the Internet. The address space announced on the Internet is an entirely separate issue. IP addresses were established as part of the development of a networking protocol called the Internet Protocol, or IP for short. This protocol was designed to allow many independent networks to interconnect or internetwork and exchange traffic. In order for such internetworks to work they need to be allocated unique IP addresses. The prerequisite for receiving globally unique IP addresses is that you have to be using IP technology and have a need to internetwork with other networks. There are several such IP internetworks that are entirely separate from the public (big I) Internet. That's where the other addresses are used and their usage is growing at about the same rate as Internet usage is growing. Think of it like Ethernet MAC addresses.
From where I sit, the perceived shortage is due to non-existant reclamation of unused resources,
It would be nice to see more reclamation and recycling of IP addresses. As near as I can tell, there is a fair bit of reclamation but it is all voluntary and ad-hoc. It isn't part of any publicly agreed process that would lead to reuse of those addresses.
All I see is that in 3-4 years we will actually have to engage our collective brains again and start getting new IP allocations from a different source.
If you agree with the Cisco IP Journal article then we have to engage our collective brains *NOW* to plan and test and be ready for the day when we need to do things differently. --Michael Dillon
Michael.Dillon@btradianz.com wrote:
Percentage of available address space announced: 38.6
You misunderstand what IP addresse are. They have nothing whatsoever to do with the Internet. The address space announced on the Internet is an entirely separate issue.
IP addresses were established as part of the development of a networking protocol called the Internet Protocol, or IP for short. This protocol was designed to allow many independent networks to interconnect or internetwork and exchange traffic. In order for such internetworks to work they need to be allocated unique IP addresses.
The prerequisite for receiving globally unique IP addresses is that you have to be using IP technology and have a need to internetwork with other networks. There are several such IP internetworks that are entirely separate from the public (big I) Internet. That's where the other addresses are used and their usage is growing at about the same rate as Internet usage is growing.
While I do not necessarily disagree with this point of view (as I work for a company who uses allocated space in such a manner), others may argue that addresses that are assigned through the Internet Assigned Numbers Authority (that's Internet with the "I") are meant for Internet, with an "I," use. As it says at the top of their web page, "Dedicated to preserving the central coordinating functions of the global Internet for the public good." Note, "global Internet." ObOnSubject: Of course, getting PI space for non-global Internet use is one of the big problems with current IPv6 allocation policy that make it difficult to start building private IPv6 networks now. -- Crist J. Clark crist.clark@globalstar.com Globalstar Communications (408) 933-4387
On Thu, Oct 13, 2005 at 09:28:19AM -0700, Crist Clark wrote: ...
While I do not necessarily disagree with this point of view (as I work for a company who uses allocated space in such a manner), others may argue that addresses that are assigned through the Internet Assigned Numbers Authority (that's Internet with the "I") are meant for Internet, with an "I," use. As it says at the top of their web page, "Dedicated to preserving the central coordinating functions of the global Internet for the public good." Note, "global Internet." ...
And what happens when a non-connected internet starts connecting to the public Internet? Even if it's through a single-layered proxy? Better to use non-conflicting IP addresses to start off with. Trying to do this with self-issued IP addresses has caused problems more than once. [Yah, "this is what RFC 1918 is for", but what if two RFC 1918 networks connect?] -- Joe Yao ----------------------------------------------------------------------- This message is not an official statement of OSIS Center policies.
I don't think so ... I recall Geoff Huston in the last APNIC indicated that this kind of actions are only going to provide a few additional time. I think the BoF should be more in the direction of "why not doing already IPv6 (from the perspective of the ISPs) ?". Delaying the inevitable don't seems the best approach to me, instead, preparing everything ahead of time, reduce the cost, which in any case is not significant. Regards, Jordi
De: Daniel Golding <dgolding@burtongroup.com> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 15:20:31 -0400 Para: Randy Bush <randy@psg.com>, "Steven M. Bellovin" <smb@cs.columbia.edu> CC: <nanog@nanog.org> Conversación: IPv6 news Asunto: Re: IPv6 news
On 10/12/05 3:13 PM, "Randy Bush" <randy@psg.com> wrote:
geoff's predictions for a very lively market in v4 space will seriously come into play.
Maybe its time to have a serious talk about IPv4 commodity trading schemes. Anyone interested in this enough to have a BOF at ARIN/NANOG?
This could extend the lifetime of the IPv4 space significantly by promoting efficient use through economic incentives, provide positive economic incentives to move to v6 when needed, and eliminate the grey market.
Proper controls could be put into place to prevent de-aggregation through utilization of the RIRs as clearing houses.
randy
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
On Wed, 12 Oct 2005, Randy Bush wrote:
if you look at the recent ipv4 burn rate of ripe and apnic especially, we run out of v4 space in about three years. this should not be surprising, as it matches what frank was saying a decade ago at ale.
so having dual stack backbones is very important. but ...
four years from now, when marissa can't get v4 space from an rir/lir and so gets v6 space, she will not be able to use 99% of the internet because no significant number of v4 end hosts will have bothered to be v6 enabled because there was no perceived market for it.
I think more likely is the scenario where Marissa would get NATed IPv4 address (NAT server at the ISP end) and one or more direct IPv6 addresses. The question would then be if Marissa is likely to use the kind of applications where the direct address would become very important to her, but so far from what I know of DSL users, most are just fine behind their home NAT firewalls and only few need direct addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space... -- William Leibzon Elan Networks william@elan.net
On Wed, 12 Oct 2005, william(at)elan.net wrote:
addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space...
Both BT and other P2P protocols are perfectly happy behind NAT. There are a few that seem to prefer that they have a non-natted address, or use some port forwarding. Those applications will just need to be fixed if it becomes a common practive of handing out NAT addresses to customers. I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons. -Sean
As I know, BT and P2P (some apps), already are using IPv6 ;-) And in 6-12 months the new Vista will start replacing XP, with IPv6 enabled by default. If you observe what is happening with XP and IPv6 NOT enabled by default, you may guess what will happen and how many apps. developers will take it seriously. Regards, Jordi
De: Sean Figgins <sean@labrats.us> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 15:56:05 -0600 (MDT) Para: <nanog@nanog.org> Asunto: Re: IPv6 news
On Wed, 12 Oct 2005, william(at)elan.net wrote:
addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space...
Both BT and other P2P protocols are perfectly happy behind NAT. There are a few that seem to prefer that they have a non-natted address, or use some port forwarding.
Those applications will just need to be fixed if it becomes a common practive of handing out NAT addresses to customers.
I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons.
-Sean
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
On Thu, Oct 13, 2005 at 01:41:26AM +0200, JORDI PALET MARTINEZ wrote:
As I know, BT and P2P (some apps), already are using IPv6 ;-)
I know of no official BitTorrent supporting IPv6... unfortunately. There were patches floating around, but to my understanding incompatible, and problems with BT servers. Otherwise I'd run an IPv6-only tracker for popular freely distributable software myself. :-) Best regards, Daniel -- CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
As I know, BT and P2P (some apps), already are using IPv6 ;-)
show flow logs please.
And in 6-12 months the new Vista will start replacing XP, with IPv6 enabled by default. If you observe what is happening with XP and IPv6 NOT enabled by default, you may guess what will happen and how many apps. developers will take it seriously.
Regards, Jordi
De: Sean Figgins <sean@labrats.us> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 15:56:05 -0600 (MDT) Para: <nanog@nanog.org> Asunto: Re: IPv6 news
On Wed, 12 Oct 2005, william(at)elan.net wrote:
addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space...
Both BT and other P2P protocols are perfectly happy behind NAT. There are a few that seem to prefer that they have a non-natted address, or use some port forwarding.
Those applications will just need to be fixed if it becomes a common practive of handing out NAT addresses to customers.
I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons.
-Sean
************************************ The IPv6 Portal: http://www.ipv6tf.org
Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com
This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
-- -------------------------------------------------------------------------- Joel Jaeggli Unix Consulting joelja@darkwing.uoregon.edu GPG Key Fingerprint: 5C6E 0104 BAF0 40B0 5BD3 C38B F000 35AB B67F 56B2
On Wed, 12 Oct 2005, Joel Jaeggli wrote:
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
As I know, BT and P2P (some apps), already are using IPv6 ;-)
show flow logs please.
It's not a flow log, but.. "Observations of IPv6 Traffic on a 6to4 Relay" (in ACM SIGCOMM CCR Internet Vital Signs special issue, January 2005), http://citeseer.ist.psu.edu/724626.html .. (section 4.4) shows that a small number of hosts (like 7-8 or so) in April 2004 used BT through our 6to4 relay. The use may or may not have gone up since, this mainly depends on whether v6 support has been included in BT. My (unverified) recollection is that BT supports v6 off-the-box in most linux distros, but I may be wrong. -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
I will be happy to show them but what I know here comes only from the paper that I've indicated a few week ago about the 6to4 relays. I just seen the same comment from the author on the list ... Anyway, the last time I attached a couple of small graphics with stats on web servers, they were filtered by the list ... Regards, Jordi
De: Joel Jaeggli <joelja@darkwing.uoregon.edu> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 16:50:56 -0700 (PDT) Para: JORDI PALET MARTINEZ <jordi.palet@consulintel.es> CC: "nanog@merit.edu" <nanog@merit.edu> Asunto: Re: IPv6 news
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
As I know, BT and P2P (some apps), already are using IPv6 ;-)
show flow logs please.
And in 6-12 months the new Vista will start replacing XP, with IPv6 enabled by default. If you observe what is happening with XP and IPv6 NOT enabled by default, you may guess what will happen and how many apps. developers will take it seriously.
Regards, Jordi
De: Sean Figgins <sean@labrats.us> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 15:56:05 -0600 (MDT) Para: <nanog@nanog.org> Asunto: Re: IPv6 news
On Wed, 12 Oct 2005, william(at)elan.net wrote:
addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space...
Both BT and other P2P protocols are perfectly happy behind NAT. There are a few that seem to prefer that they have a non-natted address, or use some port forwarding.
Those applications will just need to be fixed if it becomes a common practive of handing out NAT addresses to customers.
I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons.
-Sean
************************************ The IPv6 Portal: http://www.ipv6tf.org
Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com
This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
-- -------------------------------------------------------------------------- Joel Jaeggli Unix Consulting joelja@darkwing.uoregon.edu GPG Key Fingerprint: 5C6E 0104 BAF0 40B0 5BD3 C38B F000 35AB B67F 56B2
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
And in 6-12 months the new Vista will start replacing XP,
Will start replacing XP on new consumer-grade computers. Corporations will take another 2-4 years to switch, and other people might have upgraded to windows 98 from 3.11 by then. I think that we need to buy as much time as possible for IP, as V6 is going to be extremely painful for the consumer, and thus the consumer is not going to want to adopt it. Our jobs, as network designers and operators will be make it seemless to the consumer without forcing them to shell out a thousand or more dollars on new Windows software, and the hardware that will be required to run it on. If that is devising some sort of NAT for the large percentage of customers that don't care, then that may be the direction we need to take. I have thought for a long time that which v6 is a worthy academic persuit, customers are hardly interested in it when what they have now works. -Sean
I don't think people upgrade anymore to 98, but at least to XP (if they do now, at the end of next year will be doing to Vista). I don't think either all the corporations take so long as 2 years to upgrade. Of course, I don't have concrete logs to show on anything of this, but is not marketing just personal view based on experience with customers ;-) By the way, if we start requiring logs for any comment that we do in this list, then it may happen that the list is not so useful. I disagree also that IPv6 is painful for the consumer, on the other way around. Today they need to look into manuals for configuring STBs and other devices. Most of the time this cost a lot of troubleshooting and support to vendors and ISPs, which I know is not worth for even if charged to the customer. Consumers don't pay for IP at all, but for having things easier (not reading manuals, not needing to configure tech stuff), having more services and apps. Having more services and apps running into our networks will mean more revenue, depending on your business model (such as more free and PAY TV channels in a sat dish), and possibly because the increase in BW demand. I also see much more customers interest in IPv6 outside of NA, but may be my wrong perception, and not talking about academia. Regards, Jordi
De: Sean Figgins <sean@labrats.us> Responder a: <owner-nanog@merit.edu> Fecha: Wed, 12 Oct 2005 21:56:27 -0600 (MDT) Para: "nanog@merit.edu" <nanog@merit.edu> Asunto: Re: IPv6 news
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
And in 6-12 months the new Vista will start replacing XP,
Will start replacing XP on new consumer-grade computers. Corporations will take another 2-4 years to switch, and other people might have upgraded to windows 98 from 3.11 by then.
I think that we need to buy as much time as possible for IP, as V6 is going to be extremely painful for the consumer, and thus the consumer is not going to want to adopt it.
Our jobs, as network designers and operators will be make it seemless to the consumer without forcing them to shell out a thousand or more dollars on new Windows software, and the hardware that will be required to run it on. If that is devising some sort of NAT for the large percentage of customers that don't care, then that may be the direction we need to take.
I have thought for a long time that which v6 is a worthy academic persuit, customers are hardly interested in it when what they have now works.
-Sean
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
On Thu, 13 Oct 2005 09:20:30 +0200, JORDI PALET MARTINEZ said:
I don't think people upgrade anymore to 98, but at least to XP (if they do now, at the end of next year will be doing to Vista).
I don't think either all the corporations take so long as 2 years to upgrade.
In fact, some surveys have found that the uptake of XP in corporations has been much lower than Microsoft might like - one this summer found the number to be only '56% upgraded already'. A *lot* of sites looked at the XP 'new feature' list, the pricetag of migrating from W2K to XP, and decided it wasn't worth it. Some analysts have gone so far as to say that the continual dropping of features from Vista in order to keep the ship date from slipping even more is the single biggest threat to Microsoft - because if companies didn't think XP was worth the migration, a migration to Vista is going to need some major bang for the buck (conversely, others are saying that if there's *too* many incompatible changes in Vista, it will be a major opportunity for lower-cost migrations to Linux).
Thus spake "Sean Figgins" <sean@labrats.us>
On Thu, 13 Oct 2005, JORDI PALET MARTINEZ wrote:
And in 6-12 months the new Vista will start replacing XP,
Will start replacing XP on new consumer-grade computers. Corporations will take another 2-4 years to switch, and other people might have upgraded to windows 98 from 3.11 by then.
The companies I've worked for are usually only 6-12 months behind the latest Windows release. I'll agree that 24 motnhs was probably accurate for those moving off NT 4.0 and 95, but starting with 2k things seem to have been happening much faster since MS has been doing much better with backwards compatibility on drivers and APIs. Vista will have one big stumbling block for many -- 64bit drivers. I'm betting most corps that upgrade will initially go with the 32bit version even on 64-bit-capable PCs to keep things consistent and get the best support for legacy drivers. Or, perhaps by the time Vista comes out the vendors will have caved into the furious screams from XP64's early adopters...
I think that we need to buy as much time as possible for IP, as V6 is going to be extremely painful for the consumer, and thus the consumer is not going to want to adopt it.
Our jobs, as network designers and operators will be make it seemless to the consumer without forcing them to shell out a thousand or more dollars on new Windows software, and the hardware that will be required to run it on. If that is devising some sort of NAT for the large percentage of customers that don't care, then that may be the direction we need to take.
XP already comes with v6; all you have to do is Start->Run->"ipv6 install". Vista will just change things from "default off" to "default on". IT departments can handle the logistics of this once their network is v6-capable. The real stumbling blocks, IMHO, are: 1. In-house custom apps that don't have v6 support 2. Network hardware that needs to be replaced to handle v6 3. Stable IOS versions that support v6 4. Ops staff training on how to work with v6 5. People who rely on typing (or hard-coding) IP addresses instead of using DNS. Consumers have entirely different problems, which I'll address (no pun intended) in another subthread. S Stephen Sprunk "Those people who think they know everything CCIE #3723 are a great annoyance to those of us who do." K5SSS --Isaac Asimov
On Wed, Oct 12, 2005 at 09:56:27PM -0600, Sean Figgins wrote: ...
Our jobs, as network designers and operators will be make it seemless to the consumer without forcing them to shell out a thousand or more dollars on new Windows software, and the hardware that will be required to run it on. ... ...
Linux and the BSDs cost little and run better on older hardware. [joe ducks!] -- Joe Yao ----------------------------------------------------------------------- This message is not an official statement of OSIS Center policies.
On Wed, 12 Oct 2005, Sean Figgins wrote:
On Wed, 12 Oct 2005, william(at)elan.net wrote:
addresses. But of those "few" many are those doing P2P sharing especially with BitTorent and this application requires open port on the user end, so in fact P2P and BT may prove to be the cornerstone to getting wider use of IPv6 after we ran out of v4 space...
Both BT and other P2P protocols are perfectly happy behind NAT. There are a few that seem to prefer that they have a non-natted address, or use some port forwarding.
P2P protocols will work behind NAT only for clients. But if you want to have distributed indexes and distributed content servers (which is what P2P aims at) you need to have those who provide content to have open ports for outsiders to connect to. With NAT this is achieved by opening those specific ports which is fine for when you have home firewall but it would not work if you do not control the NAT box. But its possible to use technique where only index server has to have an open port and than require all content server clients to keep open connection to it and use that to direct them to connect to new clients requesing the data - I'm not sure if BT is doing it right now or not.
Those applications will just need to be fixed if it becomes a common practive of handing out NAT addresses to customers.
You can fix some applications but not all and when you're faced with situations that you do not even control NAT, then you have a problem.
I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons.
If its corporate system, they'd also end up using NAT (many already do). The problem would be for webhosts and ASPs who have no choice but to use real ips. -- William Leibzon Elan Networks william@elan.net
On Wed, 12 Oct 2005, william(at)elan.net wrote:
I think the bigger problem would be that of a larger company running out of RFC 1918 space, for various reasons.
If its corporate system, they'd also end up using NAT (many already do). The problem would be for webhosts and ASPs who have no choice but to use real ips.
Uh... No, I think you misunderstood. Not all 1918 space is destined to hit the Internet through NAT. Much of it's use is for devices that never, ever hit the Internet. Take, for example, STBs, modems, provisioning servers, etc. Those all tend to be customer facing, and not IT or corporate networks. The customers do not see these IPs, but systems do. Now, take a large company, such as some of the largest end-user service providers that provide not only the above, but other services as well. Add in traditional services, and you have a huge drain on 1918 space, fro things that never hit a device outside the company's network. Of course, I can not speak to what MY company does, but I can tell you that it is hard to manage. -Sean
I am told that some of the access providers are starting to deploy in the US, or at least that's what they tell us. Macs and Linux come with v6 enabled, and Longhorn will as well. So with any luck we will squeak through this one. On Oct 12, 2005, at 12:13 PM, Randy Bush wrote:
four years from now, when marissa can't get v4 space from an rir/lir and so gets v6 space, she will not be able to use 99% of the internet because no significant number of v4 end hosts will have bothered to be v6 enabled because there was no perceived market for it.
I am told that some of the access providers are starting to deploy in the US, or at least that's what they tell us. Macs and Linux come with v6 enabled, and Longhorn will as well. So with any luck we will squeak through this one.
that'll be great for the important applications such as p2p file sharing. cool! but, if you read my message, the point is that all the major hosted services will not be dual stack. half of them can't even provide well-deployed ipv4 service; try united.com. randy
On Oct 12, 2005, at 8:00 PM, Randy Bush wrote:
but, if you read my message, the point is that all the major hosted services will not be dual stack. half of them can't even provide well-deployed ipv4 service; try united.com.
That is not entirely the fault of the hosting companies.. Note that verio, he.net, towardex, and many other progressive hosting companies have been dual stack for a long time. Perhaps the services that are not able to do dual stack will vote with their wallets and either move to a company who can help them with this or at least buy better engineers. Something has to sort of make them do it though, I can't see united.com just coming up with this idea on their own. -Scott
but, if you read my message, the point is that all the major hosted services will not be dual stack. half of them can't even provide well-deployed ipv4 service; try united.com.
That is not entirely the fault of the hosting companies.. Note that verio, he.net, towardex, and many other progressive hosting companies have been dual stack for a long time. Perhaps the services that are not able to do dual stack will vote with their wallets and either move to a company who can help them with this or at least buy better engineers. Something has to sort of make them do it though, I can't see united.com just coming up with this idea on their own.
my point is that they have no incentive to do so. there are no significant v6 customers, and will likely not be until after we have blown through v4 space. this is what i mean by the bad gap. and don't you just love the suggestions of natting v6? randy
Randy Bush wrote:
and don't you just love the suggestions of natting v6?
No, but I would like to see consumer routers support rfc3068 (automatic 6to4 tunneling) by default when there is no native IPv6 access service. If we could convince manufacturers that rfc3068 is "NAT" for ipv6 they'll probably jump right on it :) - Kevin
Thus spake "Kevin Loch" <kloch@hotnic.net>
Randy Bush wrote:
and don't you just love the suggestions of natting v6?
No, but I would like to see consumer routers support rfc3068 (automatic 6to4 tunneling) by default when there is no native IPv6 access service.
If we could convince manufacturers that rfc3068 is "NAT" for ipv6 they'll probably jump right on it :)
That's probably the best suggestion yet for getting consumers on IPv6; MS has even included support for 6to4 and Teredo in XP, just waiting to be turned on. There's millions of people out there that have, with one 6to4 relay alone showing a tenfold increase in machines over the last year; that's going to be a significant number of eyeballs in another year. Unfortunately, while there's lots of great support for 6to4 in the open-source WRT software, somehow Linksys et al still haven't managed to get the feature (or even simple v6 support) into their standard releases. I have a pre-WRT Linksys which won't even support allowing 6in4/6to4 traffic through from my XP box. RFC 3068 also has another problem -- not enough relays, or at least not enough in logical locations. From my home in Texas, a traceroute shows the topologically "closest" instance of 192.88.99.1 to be in France. Nice to see that GBLX's "native" IPv6 network doesn't have any 6to4 relays in the US, and that AT&T doesn't have any at all. (Or if they do, they need serious anycast tuning) I'm not convinced that businesses will be interested in 6to4 or Teredo, though; most will want PI space and a native pipe just like they have with v4. S Stephen Sprunk "Those people who think they know everything CCIE #3723 are a great annoyance to those of us who do." K5SSS --Isaac Asimov
On Fri, 14 Oct 2005, Stephen Sprunk wrote:
RFC 3068 also has another problem -- not enough relays, or at least not enough in logical locations. From my home in Texas, a traceroute shows the topologically "closest" instance of 192.88.99.1 to be in France.
Well, anycast isn't necessarily the best way to do it. And when I lasyt checked routeviews, there weren't all that many orgs advertising the anycast special route. Yes, more should, but... <sigh> That said, even such a distant gateway would be fine for v6 *eyeballs* if organizations would voluntarily set up 6to4 outbound relays for their own v6 networks. It's as simple as setting up a route to 2002::/16 at the border with a 6to4 conversion. Only the border router itself needs to know about it; the v6 source address in the payload can continue to be native. (This still passes the security checks noted by RFC-I-can't-remember-right-now.) The great kick about this is that 6to4 eyeballs will mostly have content going *to* them, so the 6to4 relay at the organization's border will get the packets to their destination in the most efficient fashion: traveling v4 from the border to the user. This assumption is based on the typical 6to4 situation, because using 6to4 normally implies that the v4 path is the more efficient one. So, while mostly ACKs are going the slow route from the user, the big chunks of data are coming back to the user via the fast route. Example: home.duh.org is 2001:470:1f00:342::2, but if you connect to it via 6to4, the return packets will come back tunneled 6to4 directly by my border. This is doubly good for me, because my "natively addressed" v6 is tunneled anyway, so using my v6 upstream for 6to4 return traffic would mean twice the yummy tunnel slowness. Instead, my return traffic for 6to4 clients uses *zero* third party v6 tunnels. -- -- Todd Vierling <tv@duh.org> <tv@pobox.com> <todd@vierling.name>
On Fri, Oct 14, 2005 at 06:06:03PM -0400, Todd Vierling wrote:
That said, even such a distant gateway would be fine for v6 *eyeballs* if organizations would voluntarily set up 6to4 outbound relays for their own v6 networks. It's as simple as setting up a route to 2002::/16 at the border with a 6to4 conversion.
The problem is building a high performance gateway. Currently you have about the following two options: a) set up / configure a Cisco used as 6to4 gateway b) set up a dedicated host (Unix box) as 6to4 gateway Approach a) is good for only few traffic, really. Approach b) is more complex. Both approaches aren't really appealing. I'm waiting for vendor J to enable option c)... implementing 6to4 via the Tunnel PIC (or other PICs including the Tunnel PIC functionalities like Link Services PIC). It's a very simple translation/encapsulation which doesn't require any state keeping, shouldn't be a big deal. I can imagine a few larger IPv6 ISPs then suddenly implementing 6to4 gateways. :-) Regards, Daniel -- CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
On Sat, 15 Oct 2005, Daniel Roesen wrote:
It's as simple as setting up a route to 2002::/16 at the border with a 6to4 conversion.
The problem is building a high performance gateway. Currently you have about the following two options:
a) set up / configure a Cisco used as 6to4 gateway b) set up a dedicated host (Unix box) as 6to4 gateway
Approach a) is good for only few traffic, really.
<reminiscence> You know, I still barely remember when I thought IOS could do just about anything efficiently. Wow, have times changed. </reminiscence> Maybe to start -- but again, what kind of 6to4 traffic level are we expecting yet? It's the chicken and egg all over again.
Approach b) is more complex.
Yes, unfortunately.
I'm waiting for vendor J to enable option c)... implementing 6to4 via the Tunnel PIC (or other PICs including the Tunnel PIC functionalities like Link Services PIC). It's a very simple translation/encapsulation which doesn't require any state keeping, shouldn't be a big deal. I can imagine a few larger IPv6 ISPs then suddenly implementing 6to4 gateways.
The only thing that makes 6to4 more complex, compared to a plain IPIP (or GRE, or any other point-to-point vanilla tunnel protocol) tunnel is that the far-side endpoint changes based on the tunneled payload. That said, it should *not* be an unsurmountable problem -- if the demand is there. Has anyone seen if the chicken laid the hatching egg yet? -- -- Todd Vierling <tv@duh.org> <tv@pobox.com> <todd@vierling.name>
On Fri, Oct 14, 2005 at 10:45:33PM -0400, Todd Vierling wrote:
Maybe to start -- but again, what kind of 6to4 traffic level are we expecting yet?
Peak or average? Think twice before answering. :-) I'm told there are 6to4 relays seeing in excess of 100mbps. Not bursts. Can you imagine trying to handle 100mbps "internet mix" traffic process switched? :-Z Not even talking about the peaks.
The only thing that makes 6to4 more complex, compared to a plain IPIP (or GRE, or any other point-to-point vanilla tunnel protocol) tunnel is that the far-side endpoint changes based on the tunneled payload.
That's a trivial op, and the Tunnel PIC sees the L3 header anyway so can easily take it from there. But I fear that feature hasn't made it to the RFPs and otherwise "high profile much revenue anticipated" feature request lists yet, so we'll have to wait. :-( Regards, Daniel -- CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
Daniel Roesen wrote:
On Fri, Oct 14, 2005 at 10:45:33PM -0400, Todd Vierling wrote:
Maybe to start -- but again, what kind of 6to4 traffic level are we expecting yet?
Peak or average? Think twice before answering. :-)
I'm told there are 6to4 relays seeing in excess of 100mbps. Not bursts. Can you imagine trying to handle 100mbps "internet mix" traffic process switched? :-Z Not even talking about the peaks.
They may be handling 100mbps but they also have a global scope. But all the traffic statistics for 6to4 relays I've seen push several orders of magnitude *less* traffic then that. Anyone have any real numbers?
On 2005-10-15, Nicholas Suan <nsuan@nonexiste.net> wrote:
I'm told there are 6to4 relays seeing in excess of 100mbps. Not bursts. Can you imagine trying to handle 100mbps "internet mix" traffic process switched? :-Z Not even talking about the peaks. They may be handling 100mbps but they also have a global scope. But all the traffic statistics for 6to4 relays I've seen push several orders of magnitude *less* traffic then that. Anyone have any real numbers?
The 6to4 relay I run at my employer (AS29259) does around 200kbps on average and maybe short bursts of up to 5mbps. The ones doing more than 100mbps are usually the relays used outbound by large content delivery networks (the relay used by newszilla6.xs4all.nl (free binary newsserver) comes to my mind). This is totally manageable with Cisco routers regardless the constant bickering about the suboptimal Cisco (IPv6) performance which would lead you to the assumption that you can't transit a single byte of IPv6 traffic through such a box. They do work quite well these days. Bernhard
On Sat, 15 Oct 2005 00:22:15 -0500 Nicholas Suan <nsuan@nonexiste.net> wrote:
Daniel Roesen wrote:
On Fri, Oct 14, 2005 at 10:45:33PM -0400, Todd Vierling wrote:
Maybe to start -- but again, what kind of 6to4 traffic level are we expecting yet?
Peak or average? Think twice before answering. :-)
I'm told there are 6to4 relays seeing in excess of 100mbps. Not bursts. Can you imagine trying to handle 100mbps "internet mix" traffic process switched? :-Z Not even talking about the peaks.
They may be handling 100mbps but they also have a global scope. But all the traffic statistics for 6to4 relays I've seen push several orders of magnitude *less* traffic then that. Anyone have any real numbers?
According to Pekka in this paper http://www.6net.org/publications/papers/csc-6to4.pdf "The traffic of our 6to4 relay has been quite modest; in August 2004, the steady state was only about (a relatively constant) 300-500 kbit/s at 50-100 packets per second. However, quite often there are peaks to 10 Mbit/s and even beyond." Regards Marshall Eubanks
Daniel Roesen writes:
On Fri, Oct 14, 2005 at 10:45:33PM -0400, Todd Vierling wrote:
Maybe to start -- but again, what kind of 6to4 traffic level are we expecting yet?
Peak or average? Think twice before answering. :-)
I'm told there are 6to4 relays seeing in excess of 100mbps. Not bursts. Can you imagine trying to handle 100mbps "internet mix" traffic process switched? :-Z Not even talking about the peaks.
Note that not all Cisco routers use process switching for 6to4 tunnel encap/decap (which is really just IPv6-in-IPv4). Catalyst 6500/7600 OSR with PFC-3 (Sup32/Sup720) do this "in hardware". -- Simon.
On Sun, 16 Oct 2005, Simon Leinen wrote:
Note that not all Cisco routers use process switching for 6to4 tunnel encap/decap (which is really just IPv6-in-IPv4). Catalyst 6500/7600 OSR with PFC-3 (Sup32/Sup720) do this "in hardware".
And for dual-stack organizations using these at the borders, deploying a local-only 6to4 outbound relay should be easy, right? (The fact that transits should really be providing 6to4 to their downstreams notwithstanding -- it would be faster at the org's border, but if transits collectively offered 2002::/16 as standard practice, it may not be such a big deal.) -- -- Todd Vierling <tv@duh.org> <tv@pobox.com> <todd@vierling.name>
RFC 3068 also has another problem -- not enough relays, or at least not enough in logical locations. From my home in Texas, a traceroute shows
topologically "closest" instance of 192.88.99.1 to be in France. Nice to see that GBLX's "native" IPv6 network doesn't have any 6to4 relays in
the the
US, and that AT&T doesn't have any at all. (Or if they do, they need serious anycast tuning)
This seems like a problem that could be solved in the style of the CIDR report. Regular weekly reports of v6 relays and locations as seen from various major ASes. --Michael Dillon
This seems like a problem that could be solved in the style of the CIDR report. Regular weekly reports of v6 relays and locations as seen from various major ASes.
From my "tr" website I can see a few 6to4 gateways: http://tr.meta.net.nz/output/2005-10-17_22:41_192.88.99.1.png (beware, the image is extremely large, and can kill some browsers on lower end machines).
Most of my source nodes are in NZ unfortunately which limits the number of relays seen.
On Wed, Oct 12, 2005 at 08:25:05PM -0400, K. Scott Bethke wrote:
That is not entirely the fault of the hosting companies.. Note that verio, he.net, towardex, and many other progressive hosting companies have been dual stack for a long time. Perhaps the services that are not able to do dual stack will vote with their wallets and either move to a company who can help them with this or at least buy better engineers. Something has to sort of make them do it though, I can't see united.com just coming up with this idea on their own.
I can't speak for the others but he.net doesn't seem to interested in customers making use of their "dual stack" network. We looked into getting IPv6 space from them to go with our IPv4 assignments for a couple of racks of servers in one of their datacenters. They wanted to double the monthly fee for data and drop a second v6 only port to our racks, not my idea of a "dual stack network". Needless to say, we do not have native IPv6, a few of our customers that desired it are using HE's free tunnel broker service though. Michael
I can't speak for the others but he.net doesn't seem to interested in customers making use of their "dual stack" network. We looked into getting IPv6 space from them to go with our IPv4 assignments for a couple of racks of servers in one of their datacenters. They wanted to double the monthly fee for data and drop a second v6 only port to our racks, not my idea of a "dual stack network". Needless to say, we do not have native IPv6, a few of our customers that desired it are using HE's free tunnel broker service though.
It is possible that they deployed a second group of routers to do ipv6 rather than bet it all on a dual stack. Heck I've seen engineers at various places digging 7507's out of the dumpster in hopes of getting their company going on v6. HE.NET has been doing ipv6 for a long time now so it is (also) possible that their deployment pre-dates a 75% bug free IOS version for dual stack. I can't comment on any sort of pricing going on. Rumor has it that some transit and hosting companies are indeed selling IPv6 today (and surprisingly enough actually have customers buying it). -Scott
On Thu, 13 Oct 2005, Michael Greb wrote:
On Wed, Oct 12, 2005 at 08:25:05PM -0400, K. Scott Bethke wrote:
That is not entirely the fault of the hosting companies.. Note that verio, he.net, towardex, and many other progressive hosting companies have been dual stack for a long time. Perhaps the services that are not able to do dual stack will vote with their wallets and either move to a company who can help them with this or at least buy better engineers. Something has to sort of make them do it though, I can't see united.com just coming up with this idea on their own.
I can't speak for the others but he.net doesn't seem to interested in customers making use of their "dual stack" network. We looked into getting IPv6 space from them to go with our IPv4 assignments for a couple of racks of servers in one of their datacenters. They wanted to double the monthly fee for data and drop a second v6 only port to our racks, not my idea of a "dual stack network". Needless to say, we do not have native IPv6, a few of our customers that desired it are using HE's free tunnel broker service though.
(Appologies to Michael for using this comment as an opportunity to delurk. I've been biting my tongue for months through all kinds of IPv6 threads.) Hurricane's approach to IPv6 is very simple. We have the free IPv6 tunnel service and we have commercial IPv6 service. If you want commercial IPv6 service, we need to charge a fee for it in order to get the necessary funds that will eventually be required to replace all of the older infrastructure that doesn't do line rate IPv6. Hurricane's price for IPv6 is the same as IPv4 at any specific commit level. The free IPv6 tunnels are low traffic and completely automated, and the user base has a different expectations than when paying for line rate gige IPv6. In business I find that you generally should not be doing things people don't want to pay for. Exceptions to this rule are for the sake of charity, fun, or long term research and development. Right now IPv6 is nearly a zero value add on for most of the people that express interest in it. This may change in the future, however that is how it is now. There are a few interesting questions here (partially rhetorical): * Is the proper price to sell IPv6 at greater, less than, or equal to IPv4? (of course customers want the price to be less, however does it cost you more to handle it? besides it's new... shouldn't there be some kind of premium value here?) * How does the cost to provide IPv6 compare to IPv4? (can you get used equipment or do you have to buy new equipment? are there as many manufacturers? how do your options for equipment in each case compare in cost? is the equipment you are using already bought and paid for? do you have to train customers and staff? what do your existing monitoring/billing/support systems use for hardware and software?) Mike. +----------------- H U R R I C A N E - E L E C T R I C -----------------+ | Mike Leber Direct Internet Connections Voice 510 580 4100 | | Hurricane Electric Web Hosting Colocation Fax 510 580 4151 | | mleber@he.net http://www.he.net | +-----------------------------------------------------------------------+
There are a few interesting questions here (partially rhetorical):
And also: Should your company be preparing to operate v6 services at all? Popular opinion is that when the automobile was invented, all buggy manufacturers shut down. This is not true. http://www.liveryone.net/ IPv6 is one of those things that will be a market differentiator. Some people will have no need to transition and there will be a market for providing v4 services to these people. I expect that the market for v4-v6 gateway services will last for a long time. The smaller companies on this list are not faced with the same kind of technology imperatives as the larger ones. --Michael Dillon
Should your company be preparing to operate v6 services at all? Popular opinion is that when the automobile was invented, all buggy manufacturers shut down. This is not true. http://www.liveryone.net/
A buggy company founded in 1972? What kind of comparison are you trying to make? Wait 75 years after your business is gone and then start anew? sam
On Fri, 14 Oct 2005 10:45:14 CDT, "Sam Hayes Merritt, III" said:
A buggy company founded in 1972?
What kind of comparison are you trying to make? Wait 75 years after your business is gone and then start anew?
No, they were 25 years *ahead* of everybody else. Remember the .com bubble, where everybody had a buggy company with a buggy business model? :)
On Fri, 14 Oct 2005 Michael.Dillon@btradianz.com wrote:
There are a few interesting questions here (partially rhetorical):
And also:
Should your company be preparing to operate v6 services at all? Popular opinion is that when the automobile was invented, all buggy manufacturers shut down. This is not true. http://www.liveryone.net/
(as much as I hate to respond to an analogy) Yeah, with cars (the new thing) costing alot more than horses. If the same was true was true for IPv6 you would see a goldrush to provide it today. When cars were new everybody and his brother rushed to try to make cars, there were all kinds of small car companies. The money that was to be made in cars was perceived to be greater than in horses.
IPv6 is one of those things that will be a market differentiator. Some people will have no need to transition and there will be a market for providing v4 services to these people. I expect that the market for v4-v6 gateway services will last for a long time.
The smaller companies on this list are not faced with the same kind of technology imperatives as the larger ones.
Hurricane already has IPv6 in place in multiple cities and makes enough money for upgrades as needed. If the people that want to speed up IPv6 adoption would help create the expectation in the market that you will have to pay for IPv6 as its own thing then the adoption will be driven much faster. Right now the evangelists of IPv6 are like "suck it up man". We're providing the service, however you've set the expectation in the market that the value of IPv6 in addition to IPv4 is zero. If IPv6 was perceived to have any premimum value (non zero) over IPv4, it would fix most all the network build comments that you keep slapping down. Mike. +----------------- H U R R I C A N E - E L E C T R I C -----------------+ | Mike Leber Direct Internet Connections Voice 510 580 4100 | | Hurricane Electric Web Hosting Colocation Fax 510 580 4151 | | mleber@he.net http://www.he.net | +-----------------------------------------------------------------------+
On Fri, 14 Oct 2005, Mike Leber wrote:
Right now the evangelists of IPv6 are like "suck it up man". We're providing the service, however you've set the expectation in the market that the value of IPv6 in addition to IPv4 is zero.
If IPv6 was perceived to have any premimum value (non zero) over IPv4, it would fix most all the network build comments that you keep slapping down.
and you are showing that the cost to do v6 is non-zero... so that should help.
Thus spake "Mike Leber" <mleber@he.net>
On Thu, 13 Oct 2005, Michael Greb wrote:
I can't speak for the others but he.net doesn't seem to interested in customers making use of their "dual stack" network. We looked into getting IPv6 space from them to go with our IPv4 assignments for a couple of racks of servers in one of their datacenters. They wanted to double the monthly fee for data and drop a second v6 only port to our racks, not my idea of a "dual stack network". Needless to say, we do not have native IPv6, a few of our customers that desired it are using HE's free tunnel broker service though.
(Appologies to Michael for using this comment as an opportunity to delurk. I've been biting my tongue for months through all kinds of IPv6 threads.)
Hurricane's approach to IPv6 is very simple.
We have the free IPv6 tunnel service and we have commercial IPv6 service.
If you want commercial IPv6 service, we need to charge a fee for it in order to get the necessary funds that will eventually be required to replace all of the older infrastructure that doesn't do line rate IPv6. Hurricane's price for IPv6 is the same as IPv4 at any specific commit level.
It is understandable that you charge extra for a v6-enabled port due to your need to fund upgraded hardware. However, that doesn't explain why you don't deliver v4 and v6 both over the same higher-priced port. If your backbone isn't native, then a single edge box could connect to both the v4 backbone and the v6 backbone. S Stephen Sprunk "Those people who think they know everything CCIE #3723 are a great annoyance to those of us who do." K5SSS --Isaac Asimov
On Fri, 14 Oct 2005, Stephen Sprunk wrote:
It is understandable that you charge extra for a v6-enabled port due to your need to fund upgraded hardware. However, that doesn't explain why you don't deliver v4 and v6 both over the same higher-priced port.
We would be happy to do this for anybody that wants to pay for it. The earlier poster implied he didn't want to pay anything extra for IPv6. Mike. +----------------- H U R R I C A N E - E L E C T R I C -----------------+ | Mike Leber Direct Internet Connections Voice 510 580 4100 | | Hurricane Electric Web Hosting Colocation Fax 510 580 4151 | | mleber@he.net http://www.he.net | +-----------------------------------------------------------------------+
On Fri, Oct 14, 2005 at 03:54:19PM -0700, Mike Leber wrote:
On Fri, 14 Oct 2005, Stephen Sprunk wrote:
It is understandable that you charge extra for a v6-enabled port due to your need to fund upgraded hardware. However, that doesn't explain why you don't deliver v4 and v6 both over the same higher-priced port.
We would be happy to do this for anybody that wants to pay for it.
The earlier poster implied he didn't want to pay anything extra for IPv6.
You must have misread my post, I stated that we were told our bill would be double and an additional IPv6 only drop would be needed in each cabinet. Perhaps the sales person was wrong, but that is what we were told and that is what I stated in my post. Due to the price being double what it was, my employer decided it wasn't worthwhile. I imagine that we would be willing to pay a premium for native v6 but not twice what we are paying now.
Mike. Michael
I don't think users need to be charged any extra for IPv6 if it runs in the same pipe as their actual IPv4 one. Do we charge to our customers when we solve a bug or problem in our network ? IPv6 was invented to solve a "bug" in IPv4: The lack of enough addresses. Of course, now IPv6 could bring extra features, and we should take the opportunity to make new business based on that. The existence of an "unlimited" addressing space for every customer itself will allow to create new services and apps (unfortunately still to come, and that's the main issue), which we will be able to charge for. Also that will generate extra bandwidth demand, which we will also charge for. Of course, at the end is a competition problem. If some carriers/ISPs don't charge for IPv6 service, may be others will need to same if they want to stay in the market. Regards, Jordi
De: Mike Leber <mleber@he.net> Responder a: <owner-nanog@merit.edu> Fecha: Fri, 14 Oct 2005 15:54:19 -0700 (PDT) Para: Stephen Sprunk <stephen@sprunk.org> CC: North American Noise and Off-topic Gripes <nanog@merit.edu> Asunto: Re: IPv6 news
On Fri, 14 Oct 2005, Stephen Sprunk wrote:
It is understandable that you charge extra for a v6-enabled port due to your need to fund upgraded hardware. However, that doesn't explain why you don't deliver v4 and v6 both over the same higher-priced port.
We would be happy to do this for anybody that wants to pay for it.
The earlier poster implied he didn't want to pay anything extra for IPv6.
Mike.
+----------------- H U R R I C A N E - E L E C T R I C -----------------+ | Mike Leber Direct Internet Connections Voice 510 580 4100 | | Hurricane Electric Web Hosting Colocation Fax 510 580 4151 | | mleber@he.net http://www.he.net | +-----------------------------------------------------------------------+
************************************ The IPv6 Portal: http://www.ipv6tf.org Barcelona 2005 Global IPv6 Summit Information available at: http://www.ipv6-es.com This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.
Jordi, On Oct 15, 2005, at 2:09 AM, JORDI PALET MARTINEZ wrote:
I don't think users need to be charged any extra for IPv6 if it runs in the same pipe as their actual IPv4 one.
If IPv6 is tunneled through IPv4 in such a way that the ISP doesn't have to do anything special, then I suspect you wouldn't get charged extra. However, if an ISP has to run two logical networks as you do with the dual stack strategy, there will be additional costs in terms of hardware/software upgrades, technical support, troubleshooting, etc. I would think it fair that those expenses would be reimbursed somehow, perhaps with a bit extra to cover the cost of further upgrades. But then again, I don't run an ISP.
Do we charge to our customers when we solve a bug or problem in our network?
I suppose it depends on whether or not everyone agrees that the bug or problem exists and the solution proposed addresses that bug or problem.
IPv6 was invented to solve a "bug" in IPv4: The lack of enough addresses.
Actually, according to section 5.1 of RFC 1726: The initial, motivating, purpose of the IPng effort is to allow the Internet to grow beyond the size constraints imposed by the current IPv4 addressing and routing technologies. Both aspects of scaling are important. If we can't route then connecting all these hosts is worthless, but without connected hosts, there's no point in routing, so we must scale in both directions. Unfortunately, it would seem the "and routing" part was forgotten. In my opinion, the real "bug" of IPv4 was the overloading of the routing locator and the end point identifier into the same protocol field. IPv6, of course, drove into the same swamp (yelling "me too, me too", with apologies to Dave Clark) and efforts like shim6 are hacks to get around this (now obvious) problem.
Of course, now IPv6 could bring extra features, and we should take the opportunity to make new business based on that. The existence of an "unlimited" addressing space for every customer
I _really_ wish people would stop saying '"unlimited"' or 'almost infinite' when talking about IPv6 address space. It simply isn't true, even in the theoretical sense, and particularly given how address space is being allocated now. It also gives many people the wrong impression: that IPv6 addresses will mean every grain of sand in the Universe (or whatever) can have portable address space.
itself will allow to create new services and apps (unfortunately still to come, and that's the main issue), which we will be able to charge for.
Maybe it's just me, but I suspect any service that would be compelling enough, in a business sense, to drive significant IPv6 deployment would also be implementable in some way in IPv4.
Also that will generate extra bandwidth demand, which we will also charge for.
My impression is that most of the folks who provide bit pipes really want to provide enhanced services, not driving to the bottom of commodity pricing.
Of course, at the end is a competition problem. If some carriers/ ISPs don't charge for IPv6 service, may be others will need to same if they want to stay in the market.
Very true. However, if carriers/ISPs don't recover their costs, they'll probably not be around very long to compete. Rgds, -drc (speaking for myself only, of course)
In message <17229.41863.627846.668592@roam.psg.com>, Randy Bush writes:
I am told that some of the access providers are starting to deploy in the US, or at least that's what they tell us. Macs and Linux come with v6 enabled, and Longhorn will as well. So with any luck we will squeak through this one.
that'll be great for the important applications such as p2p file sharing. cool!
but, if you read my message, the point is that all the major hosted services will not be dual stack. half of them can't even provide well-deployed ipv4 service; try united.com.
Maybe ISPs should start warning them of what's coming.... --Steven M. Bellovin, http://www.cs.columbia.edu/~smb
but, if you read my message, the point is that all the major hosted services will not be dual stack. half of them can't even provide well-deployed ipv4 service; try united.com. Maybe ISPs should start warning them of what's coming....
and what do we suggest that they do, or tell them what we are doing to help? roll back the clock 12 years and inject ops clue into the ivtf v6 ivory tower? get a bigger clue by four to hammer clue into the v6 marketeers? randy
if you look at the recent ipv4 burn rate of ripe and apnic especially, we run out of v4 space in about three years.
Presumably everyone here has read the recent issue of the Cisco IP Journal which has some data to back up what Randy is saying. http://www.cisco.com/en/US/about/ac123/ac147/archived_issues/ipj_8-3/ipv4.ht...
four years from now, when marissa can't get v4 space from an rir/lir and so gets v6 space, she will not be able to use 99% of the internet because no significant number of v4 end hosts will have bothered to be v6 enabled because there was no perceived market for it.
I don't believe that this will happen. End host operating systems have had IPv6 for several years now. When the IPv4 space is exhausted, the vast majority of end-hosts will be capable of running IPv6 by simply enabling it, or installing an add-on package. Of course this does involve some effort but the magnitude of the effort is similar to patching operating systems after the latest worm/virus outbreak. I'm sure that there will be a frantic scramble, but I don't expect it to last long enough for an IPv4 black market to form. Also, I would hope that the RIRs have a formal procedure in place by this time so that IPv4 addresses can be returned and recycled. --Michael Dillon
On Thu, 13 Oct 2005 Michael.Dillon@btradianz.com wrote:
I'm sure that there will be a frantic scramble, but I don't expect it to last long enough for an IPv4 black market to form.
There's already a black market in IPv4. I've seen plenty of offers to "buy" address space through various underhanded schemes. Most take the form of creating a shell company that the space is registered to and then the buyer "acquiring" that company. -- Brandon Ross AIM: BrandonNRoss Director, Network Engineering ICQ: 2269442 Internap Skype: brandonross Yahoo: BrandonNRoss
There's already a black market in IPv4. I've seen plenty of offers to "buy" address space through various underhanded schemes. Most take the form of creating a shell company that the space is registered to and
then
the buyer "acquiring" that company.
In my opinion, the occasional underhanded deal involving shell companies does not constitute a "market". Also, I believe that ARIN is aware of this practice and that they have, in the past, taken back such addresses so the transactions were not ultimately successful. Anything involving the creation and sale of shell companies is far from the world of "commodities".
Instead of continuing with this socialistic concept that IP space is somehow owned by everyone, we should, instead, give title for IP space and allow those titles to be bought and sold freely.
Instructions for joining the ARIN PPML mailing list are found here: http://www.arin.net/mailing_lists/index.html It's the third from the top. However, I think that the notion that IP space is owned by everyone is also wrong. It isn't "space" in the sense of acreage and therefore isn't owned by anyone. We are talking about numbers here. To learn how another organization assigns numbers to people, check this page: http://www.nanpa.net/number_resource_info/co_codes.html
By my own best estimates, 50% of the allocated space today is wasted in one way or another,
This is by design. IPv4 is designed to waste numbers by allocating addresses in power-of-2 blocks. IPv6 improves things by wasting a much larger amount of numbers to ensure that every organization has their fair share of waste.
Almost all of the exhaustion problems that are on the horizon are being directly caused by inefficient use of this scarce resource, certainly all of the above is solved by a capital market.
All of the exhaustion problems are also solved by applying some magic pixie dust called IPv6 which is already available in a capital market. Just ask your company's finance people for capital to buy Cisco or Juniper boxes, then ask them for capital to buy IPv4 addresses. Which capital expenditure are they willing to release funds for? In fact, they will probably ask you to justify those new boxes and when you dig into it you will likely find that you have already paid for IPv6 boxes. --Michael Dillon
At 11:56 PM 13/10/2005, Brandon Ross wrote:
On Thu, 13 Oct 2005 Michael.Dillon@btradianz.com wrote:
I'm sure that there will be a frantic scramble, but I don't expect it to last long enough for an IPv4 black market to form.
There's already a black market in IPv4. I've seen plenty of offers to "buy" address space through various underhanded schemes. Most take the form of creating a shell company that the space is registered to and then the buyer "acquiring" that company.
Why would you call an attempt to conform to existing policy "underhand"? Seems to me that there is policy framework where policy-compliant address trading incurs certain overheads, and, according to this report, the overheads are being met. Having specified "you need to do x to move an address around", then when folk actually do 'x' its not underhand or even surprising - its what they were told was the way to do it. Perhaps the appropriate way to consider this is to consider whether the existing preconditions are rational and reasonable, or whether there is a more "user-friendly" way to interface to such address movement activities that does not involve shelf company acquisition as a side-effect.
Brandon Ross <bross@internap.com> writes:
On Thu, 13 Oct 2005 Michael.Dillon@btradianz.com wrote:
I'm sure that there will be a frantic scramble, but I don't expect it to last long enough for an IPv4 black market to form.
There's already a black market in IPv4. I've seen plenty of offers to "buy" address space through various underhanded schemes. Most take the form of creating a shell company that the space is registered to and then the buyer "acquiring" that company.
The most impressive such deal I know about took place already in 1995 when IBM bought a small norwegian company called "Norsk informasjonsteknologi" (or NIT). Most people didn't see anything interesting there. But I guess IBM noticed they "owned" 32/8. Probably don't qualify for the black market though. It was more of a grey deal. Bjørn
participants (36)
-
Bernhard Schmidt -
Bjørn Mork -
Brandon Ross -
Christopher L. Morrow -
Colm MacCarthaigh -
Crist Clark -
Daniel Golding -
Daniel Roesen -
David Conrad -
Etaoin Shrdlu -
Florian Weimer -
Fred Baker -
Geoff Huston -
Joel Jaeggli -
JORDI PALET MARTINEZ -
Joseph S D Yao -
K. Scott Bethke -
Kevin Loch -
Marshall Eubanks -
Michael Greb -
Michael.Dillon＠btradianz.com -
Mike Leber -
Nicholas Suan -
Pekka Savola -
Perry Lorier -
Randy Bush -
Richard A Steenbergen -
Sam Hayes Merritt, III -
Sean Figgins -
Simon Leinen -
Stephen Sprunk -
Steven M. Bellovin -
Todd Underwood -
Todd Vierling -
Valdis.Kletnieks＠vt.edu -
william(at)elan.net