dynamic or static IPv6 prefixes to residential customers
Hi all, I will like to know, from those deploying IPv6 services to residential customers, if you are planning to provide static or dynamic IPv6 prefixes. Just to be clear, I'm for static prefix delegation to residential customers, however I heard that some ISPs are doing dynamic delegations, the same way as is common today with IPv4. I don't thin it make sense, as the main reason for doing so in IPv4 was address exhaustion and legacy oversubscription models such as PPP/dial-up. Regards, Jordi PS: I've sent this question to NANOG, got very few answers, so trying here ... ********************************************** IPv4 is over Are you ready for the new Internet ? http://www.consulintel.es The IPv6 Company 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.
Il 27/07/2011 15:02, JORDI PALET MARTINEZ ha scritto:
Hi all,
I will like to know, from those deploying IPv6 services to residential customers, if you are planning to provide static or dynamic IPv6 prefixes.
Static, /64 at the moment but /48 planned Cheers, Ricky
Sorry, /56 planned to the customer not /48 Cheers, Ricky
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC. However, static delegations might cause IPv6 routing table explosion unless you're very careful (with dynamic IPv4 address allocation you could define address pool on BRAS and advertise just the pool prefix into the rest of the network). Ivan
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of JORDI PALET MARTINEZ Sent: Wednesday, July 27, 2011 3:02 PM To: ipv6-wg@ripe.net Subject: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
Hi all,
I will like to know, from those deploying IPv6 services to residential customers, if you are planning to provide static or dynamic IPv6 prefixes.
Just to be clear, I'm for static prefix delegation to residential customers, however I heard that some ISPs are doing dynamic delegations, the same way as is common today with IPv4.
I don't thin it make sense, as the main reason for doing so in IPv4 was address exhaustion and legacy oversubscription models such as PPP/dial-up.
Regards, Jordi
PS: I've sent this question to NANOG, got very few answers, so trying here ...
********************************************** IPv4 is over Are you ready for the new Internet ? http://www.consulintel.es The IPv6 Company
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 27 Jul 2011, at 14:45, Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
Why impossible? Tim
There's a minimum timeout of 2 hours hard-coded in the SLAAC RFC to prevent DoS attacks. Some details here: http://blog.ioshints.info/2010/12/small-site-multihoming-in-ipv6-mission.htm... Then there's the failure to detect PPPoE session loss: http://blog.ioshints.info/2010/10/dhcpv6-over-pppoe-total-disaster.html Last but definitely not least, CPEs tend to copy lease time from DHCPv6 PD to SLAAC prefix validity time (and I found no way to change that behavior in Cisco IOS), so you either overload your DHCPv6 server by using short leases or risk having delegated prefixes that will stay in the customer's CPEs for a long time. Ivan
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Tim Chown Sent: Wednesday, July 27, 2011 4:08 PM To: ipv6-wg Subject: Re: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
On 27 Jul 2011, at 14:45, Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
Why impossible?
Tim
Ivan My understanding is that while a previous prefix cannot be removed by setting the lifetime to 0 (for the reason you cited) it can be deprecated instantly by setting the preferred timer to 0. Which has the same net effect of using the new prefix. -éric
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Ivan Pepelnjak Sent: mercredi 27 juillet 2011 10:26 To: 'Tim Chown' Cc: ipv6-wg@ripe.net Subject: RE: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
There's a minimum timeout of 2 hours hard-coded in the SLAAC RFC to prevent DoS attacks. Some details here:
http://blog.ioshints.info/2010/12/small-site-multihoming-in-ipv6-mission.htm...
Then there's the failure to detect PPPoE session loss:
http://blog.ioshints.info/2010/10/dhcpv6-over-pppoe-total-disaster.html
Last but definitely not least, CPEs tend to copy lease time from DHCPv6 PD to SLAAC prefix validity time (and I found no way to change that behavior in Cisco IOS), so you either overload your DHCPv6 server by using short leases or risk having delegated prefixes that will stay in the customer's CPEs for a long time.
Ivan
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Tim Chown Sent: Wednesday, July 27, 2011 4:08 PM To: ipv6-wg Subject: Re: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
On 27 Jul 2011, at 14:45, Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
Why impossible?
Tim
Yes, that's the RFC4192 method, which we have used for an enterprise (partial) renumber. I think Ivan's question is more about the CPE behaviour if there's a no-flag-day renumbering event? If the renumbering is planned, it ought to be possible to introduce the new prefix, turn down the preferred timer on the old one, run with both for a while, then remove the old prefix. If there's something the implementation or standards stopping that,what is it, and how do we fix it? I think something similar is supported in IOS if you use 6to4 and your CPE's IPv4 address changes - maybe Eric can comment on that. Not that I'd suggest using 6to4 any more ;) Tim On 27 Jul 2011, at 16:07, Eric Vyncke (evyncke) wrote:
Ivan
My understanding is that while a previous prefix cannot be removed by setting the lifetime to 0 (for the reason you cited) it can be deprecated instantly by setting the preferred timer to 0. Which has the same net effect of using the new prefix.
-éric
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Ivan Pepelnjak Sent: mercredi 27 juillet 2011 10:26 To: 'Tim Chown' Cc: ipv6-wg@ripe.net Subject: RE: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
There's a minimum timeout of 2 hours hard-coded in the SLAAC RFC to prevent DoS attacks. Some details here:
http://blog.ioshints.info/2010/12/small-site-multihoming-in-ipv6-mission.htm...
Then there's the failure to detect PPPoE session loss:
http://blog.ioshints.info/2010/10/dhcpv6-over-pppoe-total-disaster.html
Last but definitely not least, CPEs tend to copy lease time from DHCPv6 PD to SLAAC prefix validity time (and I found no way to change that behavior in Cisco IOS), so you either overload your DHCPv6 server by using short leases or risk having delegated prefixes that will stay in the customer's CPEs for a long time.
Ivan
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Tim Chown Sent: Wednesday, July 27, 2011 4:08 PM To: ipv6-wg Subject: Re: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
On 27 Jul 2011, at 14:45, Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
Why impossible?
Tim
I think something similar is supported in IOS if you use 6to4 and your CPE's IPv4 address changes - maybe Eric can comment on that. Not that I'd suggest using 6to4 any more ;)
Me neither (about using 6to4)... The advertized prefix in 'inside LAN' RA will change when the 'external WAN' IPv4 address change. But, I have no clue whether it will include the previous prefix with a preferred set to 0. Could be done by TCL & EEM. -éric
Hi Ivan /et. al/, On top of these problems add the incorrect RFC behavior of MS (i.e. Win 7) clients when dealing prefixes whose lifetime is set to 0) and you end up puzzled. Should I deploy IPv6 on on top of PPPoE ? or directly on ethernet ifces ? should I ask my CPE vendors to break the strict RFC behavior ( for MS clients)? Cheers, Dimitris On 27/7/2011 5:26 μμ, Ivan Pepelnjak wrote:
There's a minimum timeout of 2 hours hard-coded in the SLAAC RFC to prevent DoS attacks. Some details here:
http://blog.ioshints.info/2010/12/small-site-multihoming-in-ipv6-mission.htm...
Then there's the failure to detect PPPoE session loss:
http://blog.ioshints.info/2010/10/dhcpv6-over-pppoe-total-disaster.html
Last but definitely not least, CPEs tend to copy lease time from DHCPv6 PD to SLAAC prefix validity time (and I found no way to change that behavior in Cisco IOS), so you either overload your DHCPv6 server by using short leases or risk having delegated prefixes that will stay in the customer's CPEs for a long time.
Ivan
-----Original Message----- From: ipv6-wg-admin@ripe.net [mailto:ipv6-wg-admin@ripe.net] On Behalf Of Tim Chown Sent: Wednesday, July 27, 2011 4:08 PM To: ipv6-wg Subject: Re: [ipv6-wg] dynamic or static IPv6 prefixes to residential customers
On 27 Jul 2011, at 14:45, Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
Why impossible?
Tim
-- ------------------------ Dimitrios K. Kalogeras Electrical Engineer Ph.D. Network Engineer Netmode NTUA Lab _____________________________________ skype: aweboy voice: +30-210-772 1448 fax: +30-210-772 1866 e-mail: D.Kalogeras@noc.ntua.gr
On Wednesday 27 July 2011 15:45:05 Ivan Pepelnjak wrote:
Unfortunately you have to do static prefix delegation because it's impossible to renumber the customer's inside LAN within a reasonable time interval with today's state of IPv6 SLAAC.
However, static delegations might cause IPv6 routing table explosion unless
Hi, My policy is to give semi-static prefixes, that is, prefixes that are statically associated to the customer for what concerns radius & such but have NO written GUARANTEE to remain permamently assigned to the user. So, if I need to renumber I can do it with little worries, maybe a notice to the user if I'm really good. The customer knows that he has to consider the prefix as not assigned to him, thus, if he wants to run a service on those addresses he has to be prepared to use a dynamic DNS service or such. Bye,
On Wednesday 27 July 2011 18:47:19 Daniele Orlandi wrote:
My policy is to give semi-static prefixes, that is, prefixes that are statically associated to the customer for what concerns radius & such but have NO written GUARANTEE to remain permamently assigned to the user.
I forgot to add that if the customer wants such guarantee he may obtain it by paying an extra fee. Bye, -- Daniele "Vihai" Orlandi Bieco Illuminista #184213
On 7/27/11 3:02 PM, JORDI PALET MARTINEZ wrote:
Hi all,
I will like to know, from those deploying IPv6 services to residential customers, if you are planning to provide static or dynamic IPv6 prefixes.
static. no questions asked :) people should stop thinking with IPv4 minds, specially when designing big IPv6 networks (in multiple wrong and horrible IPv4 ways) Cheers, Jan
On 07/27/2011 10:01 PM, Jan Zorz @ go6.si wrote:
On 7/27/11 3:02 PM, JORDI PALET MARTINEZ wrote:
Hi all,
I will like to know, from those deploying IPv6 services to residential customers, if you are planning to provide static or dynamic IPv6 prefixes.
static. no questions asked :)
people should stop thinking with IPv4 minds, specially when designing big IPv6 networks (in multiple wrong and horrible IPv4 ways)
Cheers, Jan
In our ipv6 broadband trials, we've tested both static and dynamic scenarios. We're gearing towards a static model that is not "roaming", i.e the users will somehow be bound by their location. This way, we'll be able to aggregate customer ranges per terminating device. cheers, Yannis p.s.: I need to check the data but I'm pretty sure that we didn't experience major delays in our dynamic trials when a new prefix was assigned to an end user.
On 8/23/11 3:50 PM, Yannis Nikolopoulos wrote:
In our ipv6 broadband trials, we've tested both static and dynamic scenarios. We're gearing towards a static model that is not "roaming", i.e the users will somehow be bound by their location. This way, we'll be able to aggregate customer ranges per terminating device.
Good choice imho...
p.s.: I need to check the data but I'm pretty sure that we didn't experience major delays in our dynamic trials when a new prefix was assigned to an end user.
CPE is not the problem, SLAAC is. CPE will change the prefix on LAN side, but hosts might behave weirdly for some time (even if you start sending RA packets with 0 lifetime...) Cheers, Jan
participants (9)
-
Daniele Orlandi -
Dimitris Kalogeras -
Eric Vyncke (evyncke) -
Ivan Pepelnjak -
Jan Zorz ＠ go6.si -
JORDI PALET MARTINEZ -
Riccardo Losselli -
Tim Chown -
Yannis Nikolopoulos