IPv4 Exhaustion

derfderfley said:
Basically UK ISP's that haven't, need to start planning now, and start implementation by the middle of the year. Now this will probably be a two pronged attack;
1) Implementation of "carrier grade" NAT to stretch out the last of the v4 space.
Wave goodbye to more that a single static public IP address as a residential customer, and then get ready to be put behind NAT as the business customers that pay big money for their connections will be getting your public IP addresses.

2) Rolling out dual stack connections to all users, residential customer will probably get a /64 IPv6 range, most businesses will be given a /48.
Click to expand...

I'm not sure many standard wired ISPs will need to implement Carrier grade NAT. We are not far off market saturation for broadband. Most people move from one ISP to another, so overall the number of IP addresses needed for wired broadband remains fairly constant. The main driver, and likely location for LSN, is in the wireless space, Mobile phones that now provide Internet access instead of just telephony and sms. Indeed many carriers already have LSN in the wireless network. Have you ever looked at the IP address on your phone. Mine has an address from the 10/8 block, so somewhere along the line its being NAT'd. Similar with O2's network, you have to use a different ASN to get a public address on your phone.
 
sjcf1 said:
This is going to sound stupid. But what happens when we move to IPv6. Will there be 2 seperate internets? One for Ipv4 and one for IPv6?
Click to expand...

Yup. Nail head. Spot on.

There are methods of translation between the two, but ideally you'd just connect to both natively, (at the same time over one physical connection) and one day in the future you and your ISP will realise that very little traffic is going over IPv4, and will then let it die. It will undoubtedly be a very long, very slow, and hideously drawn out death.
 
It's all about the end devices. Pretty much no consumer grade router supports it currently, and it's quite a lot of effort for ISP's to implement dual stack when all but a handful of users won't be able to take advantage of it.
 
I remember a presentation that a chap from Cisco gave at a small conference I was at a few months ago, he was saying before they went IPv6 they had over 30million IPv4 addresses (I've got the breakdown as to how that was made up, including 2 x /8 etc etc).

Now they are IPv6 they have 2 x IPv6/48, 3 x IPv6/32 and 1 x IPv6/24, that's a lot of ips!
 
daz said:
Reclaiming old/legacy space plus forcing residential ISPs to use NAT will extend IPv4 usage significantly. I'd be surprised if it wasn't still predominant in 5 years.
Click to expand...

As I said, the allocation rate between 2008-2010 was 10 per year. For us to solve the exhaustion problem by this method alone, we'd have to free up 50 entire /8 blocks right now. And that's optimistically assuming our IP usage rate won't increase from 2008-2010 values (it really will).

Forcing companies to do things is something I'm not a particularly big fan of, and I doubt the ISPs would be happy having to put up with the massive extra cost of running kit that can support a large NAT rollout (NAT isn't cheap, and doesn't scale well). Not to mention the fact that it would only ever be a temporary solution.

Much easier to just solve the problem correctly. Like bidredshark says, the problem is really a peripheral device problem - decent networks have had ipv6 ready to go for some time now.
 
You must log in or register to reply here.
Back
Top Bottom