|
IPv6 tunnel on 4G
![[us]](/s/countries/us.gif) Shadow Hawkins on Sunday, 14 August 2011 02:51:32
Hi,
I would like to start by saying that I find sixxs connectivity to be more reliable that another broker that shall remain nameless, I am actually having difficulty with dropped packets and connections to my email host over there but consistently reliable results here.
I would like to request a second tunnel to use when away from home. I have 4G mobile broadband and would like to establish a tunnel for my laptop away from home. The problem is that this tunnel would down more than it would be up, does the ISK penalty apply to a second tunnel for this kind of application?
My email host serves on IPv6 and I would also like to show others (Fellow students, I'm currently in school for network administration) what can be done but at the same time I don't want to destroy my (ISK) credit rating either. :)
Also I've been reluctant to request a subnet just because a /48 is scary BIG, but could a single /48 be shared on 2 different tunnels? Would I have the authority over the subnet to allocate a /56 to each tunnel?
Thanks
Paul
IPv6 tunnel on 4G
Only static tunnels have penalties, that is why they get credits, if up, on a weekly basis. AYIYA and heartbeat tunnels don't have penalties and only gain credits, if up, on a bi-weekly basis.
As you said '4G', most likely the best tunnel is AYIYA as that will punch through NATs and is dynamic address aware.
Also I've been reluctant to request a subnet just because a /48 is scary BIG, but could a single /48 be shared on 2 different tunnels?
Can't be done, the /48 is routed down to one tunnel only.
IPv6 tunnel on 4G
Shadow Hawkins on Sunday, 14 August 2011 14:15:08
Thanks Jeroen,
Somewhere in the back of my head I knew that I wouldn't be able to split the subnet that way. My laptop will really be the only device using the 4G connection most of the time and it won't serve so I'll just use the second tunnel as a single host for connectivity to IPv6 land. I'm having enough trouble justifying the need for 1 /48 I'd feel totally guilty about taking 2! I know there's plenty of address space but still, how am I going to use 130,000 networks each capable of host ALL of IPv4's address space many times over. After all isn't that how we first got into trouble with IPv4, giving class A networks to entities that only used a small portion of the available space and left the rest to rot?
Thanks for your help.
Paul
IPv6 tunnel on 4G
A /48 is only 65536 /64 networks actually, thus not too much, but it should be more than enough for anybody except the largest enterprises in the world.
As for 'running out', currently IANA is assigning out of 2000::/3 only, if that ever runs full, then we still have 7 other /3's left to use with a different addressing policy which would make the address space not run out that quickly.
That said, the bigger problem becomes the routing tables, not the amount of address space given to endsites.
IPv6 tunnel on 4G
Shadow Hawkins on Sunday, 14 August 2011 15:19:32
Here I have to plead ignorance, I start studying routers in about 3 weeks. Being an end user my experience has been with very simple DSL routers. I have no understanding of routing tables, yet........
As far as my addressing analogy, I meant 130,000 to be 2 /48's and while I understand that IPv6 addresses hosts differently than IPv4 a /64 can still address as many hosts as the entire IPv4 space squared can't it? These numbers are just so big I try to put them in some sort of perspective.
Thanks again
Paul
IPv6 tunnel on 4G
I have no understanding of routing tables, yet........
In short: IPv4 currently has ~350.000 entries globally, and routers already have a hard time keeping up with that as those entries all might change and you might have that table several times from multiple transit providers.
As such, if that table grows much more, that is the first limit that will be hit.
while I understand that IPv6 addresses hosts differently than IPv4
It is mostly just a bigger number, nothing else
a /64 can still address as many hosts as the entire IPv4 space squared can't it?
Theoretically yes, but technically you will have run out of CAM table space a long long time before that ;)
Most cheap switches come with tables which hold only 8192 MAC addresses, if you have a really pricy one you might get a 128k if really lucky, but much more than that I have never seen ;)
IPv6 subnetting is quite different from IPv4, instead of picking the right number of bits to fit exactly the number of hosts that you want to have in that subnet, you simply put a /64 on that Ethernet segment and be done with it.
If you have then a router with multiple interfaces, each interface gets a /64 presto. Nice, clean and easy.
And that is where the idea of /48 per end-site comes from, as it gives full flexibility to route a /56 to each building in that organisation and subnet again from there.
IPv6 tunnel on 4G
Shadow Hawkins on Monday, 15 August 2011 00:37:04
So if I understand you right, if subnets are issued in smaller than /48 increments then the address space will last longer but long before we even make a dent in the address supply, current router technology will be swamped just trying to keep track of it all. Is there an effort being made to address that issue? I'm just getting started on a network administration degree but unfortunately IPv6 will only be mentioned in the final classes, apparently there isn't much concern yet......
I like the split a /48 into /56's one for each building but apparently with my access being limited to tunnels I can't do that. Of course I suppose it would help if I wasn't terminating one of the tunnels on a mobile device. Is there any way to have my /48 split at the Sixxs end and a /56 mapped to each tunnel? I'm a student trying to learn how to work with IPv6 and just feel greedy asking for 2 /48's for a dozen hosts.
I'm getting it slowly, I really appreciate the opportunity to work it all out and get some understanding before "crunch" time.
Thanks again for all your help, I wish my school saw the value of adding IPv6 to the curriculum.
Paul
IPv6 tunnel on 4G
So if I understand you right, if subnets are issued in smaller than /48 increments then the address space will last longer but long before we even make a dent in the address supply, current router technology will be swamped just trying to keep track of it all.
Correct. (though there is some stretching available possible to a million routes, it is hard to get there).
There are 2^29 /32's in 2000::/3 alone though which will never fit in there.
I like the split a /48 into /56's one for each building but apparently with my access being limited to tunnels I can't do that.
Well, you could build tunnels on your side of the tunnel and then split it. But the PoP won't do this for you.
Is there any way to have my /48 split at the Sixxs end and a /56 mapped to each tunnel?
No, the PoP won't do this.
I'm a student trying to learn how to work with IPv6 and just feel greedy asking for 2 /48's for a dozen hosts.
Nothing greedy.
|
![[ch]](/s/countries/ch.gif)
on Sunday, 14 August 2011 11:34:02
Posting is only allowed when you are