brad
join:2007-09-06
Etobicoke, ON | reply to SimplePanda
Re: IPv6 beta said by SimplePanda:Why they give a /56 for the LAN side I don't understand though.
So that customers can have more than one segment on the LAN side. /56's are cheap.
said by SimplePanda:Realistically all you really need is a /64 for the WAN interface and a /64 for the LAN. I'm happy having the /56 but it seems like the plentiful nature of v6 addresses has ISP's giving out insane numbers of them to customers.
Straight up auto configuration with DHCPv6 only ever assigns on the first /64 of the assigned /56 (without specific configuration otherwise) so the vast majority of people will never use more than a /64 anyways.
There's nothing insane about it. A /64 on the LAN side assumes only a single segment. /60's would probably be better but /56's are fine as well. |
|
| I think there was some suggestion in one of the RFCs that residential customers should get a /56 and commercial accounts should get a /48. Or some sort of industry defacto type standard. Anyway, that seems to be how it's shaking out.
Technically, you don't need a /64 on the WAN side, either. I've brought up my TekSavvy link using the /64 for my LAN, and using the link-local addresses for the WAN gateway. Works fine. Just means that if you want to remotely log into your router, you're technically going through the WAN to the LAN to log in. I think the /64 on the WAN side is really just an artifact of IPv4 thinking. But they're cheap, and if it makes people feel more comfortable about the deployment, then it's worthwhile. |
|
|
|
Reviews:
 ·linode
 ·Cogeco Cable
 ·Bell Sympatico
1 edit | said by Mersault:I think there was some suggestion in one of the RFCs that residential customers should get a /56 and commercial accounts should get a /48. Or some sort of industry defacto type standard. Anyway, that seems to be how it's shaking out.
That's RFC6177. It clearly states something that simultaneously proves and disproves my point:
This document moves away from the previous recommendation that a single default assignment size (e.g., a /48) makes sense for all end sites in the general case. End sites come in different shapes and sizes, and a one-size-fits-all approach is not necessary or appropriate. The rest of the RFC seems to suggest that, although it's not one-size-fits-all, it should be between (& not including) /64 and /48. So there's nothing wrong here, I guess. Personally, I don't see the need for anything as big as a /56, but I guess some people will have a use for it.
I guess what I'm still wondering is why there are two separate subnets, and why there can't just be one /56 or one /64 (or one /whatever) (things my CCNA didn't teach me). |
|
| said by squircle:I guess what I'm still wondering is why there are two separate subnets, and why there can't just be one /56 or one /64 (or one /whatever) (things my CCNA didn't teach me).
I think it probably just has to do with the way the PPPoE service works with TekSavvy. I recall configuring a PPPoE daemon, and basically you assign an address to an endpoint (in the IPv6 world, this is a /64), and then you would separately assign any routes that you wanted to route in addition. Hence in the IPv4 world you get a static IP, and then a completely different /29, if you order them from TekSavvy. Same general deal with IPv6 (this is conjecture, but it's what I suspect). |
|
| reply to squircle
Also, check out this news from Google. |
|
brad
join:2007-09-06
Etobicoke, ON | »www.worldipv6launch.org/
Google properties, Facebook, the remaining Microsoft properties, Yahoo, Cisco to name a few. Akamai, Limelight and a few other CDNs going full v6 production was/is a blocker for thousands of other sites going v6. This will help to push IPv6 much further along having so many large sites with AAAA records and no whitelisting games or special URLs.
IMO this should be a challenge for TSI to be much further along by June 6th. Have full production v6 for both ON/QC and AB/BC DSL customers by then. There is no shortage of CPE gear from D-Link/Netgear/Linksys and some other vendors as well as third-party firmware options to allow for v6 over DSL. |
|
brad
join:2007-09-06
Etobicoke, ON | reply to squircle
said by squircle:Personally, I don't see the need for anything as big as a /56, but I guess some people will have a use for it.
A /60 for consumer connections which allows for up to 16 /64's would be a good default, maybe allow for an opt-in for a /56. Business connections are good with a /56 by default and opt-in for a /48. |
|
brad
join:2007-09-06
Etobicoke, ON | reply to squircle
said by squircle:said by SimplePanda:Straight up auto configuration with DHCPv6 only ever assigns on the first /64 of the assigned /56 (without specific configuration otherwise) so the vast majority of people will never use more than a /64 anyways.
I think you meant SLAAC, but that's my exact point. And I know the whole mantra with IPv6 is that we don't really need to conserve addresses since there are so damn many of them (unlike IPv4), but I like my powers of two, and... He could have meant DHCPv6 too but both issues are router implementation specific issues and not inherent to the protocols. |
|
Reviews:
·TekSavvy DSL
·Verizon Online DSL
·Primus Telecommu..
 ·BroadVoice
·Bell Sympatico
| reply to TSI Gabe
Hello:
I am not sure what the 'TekSavvy Direct' forum is, so I will put my request here.
I am a new Teksavvy VDSL customer, switching over from Primus DSL.
I just received a IPv6 capable router, that will run the special Tomato
firmware, and I would like to request a v6 PPPOE login.
I am planning on running a network, so I need both: /64 AND a /56 entry.
Please let me know what I need to do to get this login.
thanks! Stefan in Montreal
--
Sung to the old New England Telephone Ad Jingle: " We're the only one New Eng-Lund, Hell-Ri-zon Tele-phonnnnne! " |
|
| I believe you're looking for this: » TekSavvy Direct |
|
MPower
join:2008-11-17
San Marcos, CA | So has there been any changes to Teksavvy's IPv6 implementation after the JunosE update the other night? |
|
 TSI GabePremium,VIP
join:2007-01-03
Chatham, ON
kudos:2 | not yet, the upgrade was in order to move forward with making IPv6 available to everyone, but doing so will take time, it's not just a matter of doing a software upgrade.
The goal at the moment is to be ready before the IPv6 world launch day. (6th of June) |
|
brad
join:2007-09-06
Etobicoke, ON | said by TSI Gabe:not yet, the upgrade was in order to move forward with making IPv6 available to everyone, but doing so will take time, it's not just a matter of doing a software upgrade.
The goal at the moment is to be ready before the IPv6 world launch day. (6th of June)
Thanks for the work to date.
I see Cisco finally has added IPv6 support for IronPort. What
is the status with regard to IPv6 capability for the rest of your mail infrastructure?
At the moment what is holding you back from adding a AAAA record for the TSI web-site to start off with? |
|
| reply to squircle
said by squircle:said by SimplePanda:Straight up auto configuration with DHCPv6 only ever assigns on the first /64 of the assigned /56 (without specific configuration otherwise) so the vast majority of people will never use more than a /64 anyways.
I think you meant SLAAC, but that's my exact point. And I know the whole mantra with IPv6 is that we don't really need to conserve addresses since there are so damn many of them (unlike IPv4), but I like my powers of two, and... [att=1] I did mean SLAAC, yes. Specifically, what I meant is that the out of the box configuration most people will be using, at least right now, only assigns on the first /64 of the /56 anyways (Tomato, for example). |
|
1 edit | reply to brad
said by brad:said by SimplePanda:Why they give a /56 for the LAN side I don't understand though.
So that customers can have more than one segment on the LAN side. /56's are cheap. said by SimplePanda:Realistically all you really need is a /64 for the WAN interface and a /64 for the LAN. I'm happy having the /56 but it seems like the plentiful nature of v6 addresses has ISP's giving out insane numbers of them to customers.
Straight up auto configuration with DHCPv6 only ever assigns on the first /64 of the assigned /56 (without specific configuration otherwise) so the vast majority of people will never use more than a /64 anyways.
There's nothing insane about it. A /64 on the LAN side assumes only a single segment. /60's would probably be better but /56's are fine as well. Not really sure I can imagine the end-user scenario for 99.99% of residential customers where multiple subnets are required... and realistically this won't change anytime soon. Even as people add devices to their LAN most people don't even consider what IP is never mind bothering to neatly segment their network.
Perhaps 'insane' is the wrong word. "Unnecessary in almost all situations" is probably a better description. |
|
brad
join:2007-09-06
Etobicoke, ON
1 edit | said by SimplePanda:Not really sure I can imagine the end-user scenario for 99.99% of residential customers where multiple subnets are required... and realistically this won't change anytime soon. Even as people add devices to their LAN most people don't even consider what IP is never mind bothering to neatly segment their network.
Perhaps 'insane' is the wring word. "Unnecessary in almost all situations" is probably a better description.
The 99.99% is exaggerated and they still need to cover users who need the address space. One size allocation though could cover everyones needs though with a /60 or even a /56 is fine too. There are more than 0.01% users using the v6 beta service now would require greater than one /64 never mind when they rollout to their whole customer base. Makes it easier for TSI and the users. |
|
1 edit | said by brad:said by SimplePanda:Not really sure I can imagine the end-user scenario for 99.99% of residential customers where multiple subnets are required... and realistically this won't change anytime soon. Even as people add devices to their LAN most people don't even consider what IP is never mind bothering to neatly segment their network.
Perhaps 'insane' is the wring word. "Unnecessary in almost all situations" is probably a better description.
The 99.99% is exaggerated and they still need to cover users who need the address space. One size allocation though could cover everyones needs though with a /60 or even a /56 is fine too. There are more than 0.01% users using the v6 beta service now would require greater than one /64 never mind when they rollout to their whole customer base. Makes it easier for TSI and the users. I was really more trying to say "the vast majority". There aren't many residential DSL customers in the customer base at large who "require" 256 subnets (or even 16).
I'd also suggest that rolling it out to the whole customer base will decrease the percentage of people who need more than a single /64, rather than increase it. People who want multiple subnets are likely already in the beta.
I suppose it's all moot in that TSI -can- hand out /56's en masse if they so choose. It just seems like wasteful overallocation to me. |
|
| said by SimplePanda:It just seems like wasteful overallocation to me.
It's a gross overallocation for the way we consume the internet *at present*. Can you really safely say the same for 15 years from now? Sure, you only have a few computers at home, but once integration is better and the internet embeds itself ever deeper into your house, it's going to change. A subnet for all my lighting fixtures makes sense, for instance.
Also, what if the one-IP-per-device template we use right now starts to break down? Lots of communication within a device happens over the loopback interface, I wouldn't be surprised to find that as different devices within the home need to talk to each other that they don't just start giving IPs to parts of devices. The DAC in your home audio system will have it's own IP, and your remote can talk directly to just that component.
With IPs no longer a constraining resource, I'm interested to see some true innovation in networking in the home. |
|
Reviews:
·linode
·Cogeco Cable
·Bell Sympatico
| said by Mersault:I wouldn't be surprised to find that as different devices within the home need to talk to each other that they don't just start giving IPs to parts of devices. But that's what fe80::/10 is for... |
|
rev
join:2011-12-14
Toronto, ON | said by squircle:But that's what fe80::/10 is for...
»tools.ietf.org/html/rfc3879
"Deprecating Site Local Addresses"
Heard there was a vote recently (jan 2012) that was in favour of it, I read it in passing and am too lazy to get a citation, so grain of salt please.
I for one, do not want my smart shelves on a site-local address. |
|
