Skip Navigation


Hyperoptic adds £5 charge for static IP as CGNAT arrives
Tuesday 27 September 2016 10:03:28 by Andrew Ferguson

ISPreview has lead this morning with the news that Hyperoptic is introducing a £5/m charge for those who want a static IP address, and where in the past IP addresses were on a long term lease or dynamic, dynamic IP addresses will be the order of the day and more importantly these will be via carrier grade NAT (CGNAT).

Hyperoptic is far from being the first provider to dabble with CGNAT, as BT Consumer has done so for some of its entry level products, and mobile providers commonly use it. The likely reason for the adoption is that IPv4 addresses are getting harder to find and the cost of buying spare blocks from other providers is increasing, so making maximum use of the resources you already have is the order of the day. The vast majority are not likely to spot any major problems, but there will be side-effects e.g. one IP address blocked due to abuse might affect hundreds of people. The ultimate solution is the roll-out of IPv6 but as with many providers this is taking longer than expected and as the adoption of IPv6 by services over the Internet is also lagging even if you have IPv6 you are still forced back to IPv4 for some sites/services.

IPv4 exhaustion is not a symptom of a failing UK Digital Economy but it is actually a global problem. The trade in IPv4 blocks generally only becomes visible when overseas blocks appear in the UK but services such as geo-blocking for BBC iPlayer have yet to be updated.

Time will tell how well the Hyperoptic implementation of CGNAT is implemented, change at broadband providers is never popular unless it is a price cut or a truly free speed boost.

Comments

Posted by Going_Digital 2 months ago
The lack of adoption of IPv6 by ISPs is holding things back. A solution of a proper IPv6 address along with NAT IPv4 as fallback seems the most sensible option to me. ISPs need to be helping to push people towards IPv6 not shying away from it.
Posted by TheEulerID 2 months ago
If it was as simple as ISPs not supporting IPv6, then the issue would be moderately easy to resolve. The problem is that there are gazillions of services and devices out there running IPv4 which would require massive efforts to change and little incentive to do so.
Running dual stacks doesn't reduce demand on IPv4 addresses and other co-existence methods come with drawbacks. The IPv6 designers neglected to make the two protocols interwork properly.
Posted by TheEulerID 2 months ago
Just to back this up, here's an admission that the designers of IPv6 screwed up over the interworking issue. Something that was blindingly obvious to those of use who have struggled with the principles of IT migration.

If it had been designed properly, the migration process would have been tricky, but doable. As it is, it's almost prohibitively difficult.

http://www.networkworld.com/article/2265836/lan-wan/biggest-mistake-for-ipv6--it-s-not-backwards-compatible--developers-admit.html
Posted by matthewsteeples 2 months ago
@TheEulerID I disagree. There's no way that you could have made it backwards compatible without compromising so many of the benefits of it. You'll notice that while the article cites that as a reason for slow adoption, it doesn't state how they'd have got around it. I'm interested to hear what your solution would be though :)
Posted by matthewsteeples 2 months ago
...There may be devices and services out there running only IPv4, but what is their shelf life? Every new device (should) support IPv6 out of the box, and every current OS supports IPv6 which means there's no reason that every service out there can't be supporting it. IPv4 can hang around for legacy devices, but unless there's over 2 billion of them there isn't going to be a shortage of addresses!
Posted by TheEulerID 2 months ago
@matthewsteeple

Many parts of the IT industry have been used to making new and old interoperate. As for the shelf life of equipment, then that can be very long indeed. But the issue isn't even just about hardware. Because of the low level way that TCP/IP stacks interwork with software, there will be huge numbers of applications that simply won't work with IPv6 without being updated, and that can be major work (sometimes it means new apps). Getting funds to update legacy software is always a problem in organisations.
Posted by TheEulerID 2 months ago
As far as solutions so, it would have been possible to build the interoperability into the new stack. Several decisions were made in IPV6 which seem to have valued purity over practicality. That there are still development efforts going on to try and make IPV4/6 interoperability and transitioning less painful speaks volumes. It's now 18 years since IPv6 was finalised.
Posted by CarlThomas 2 months ago
Wow these guys really don't have that many IPv4 addresses available given they wanted to pass 500,000 premises.
Posted by matthewsteeples 2 months ago
@TheEulerID I think the problem is that you simply can't make new and old interoperate without either changing the old or having massive amounts of infrastructure in place. Neither of these are practical on internet scale developments. The applications are less of an issue in my eyes too (as a software developer) as it's mostly changing data structures so an update would handle it.
Posted by matthewsteeples 2 months ago
"Legacy" is definitely the word I'd use for it. If you have something that only talks IPv4 then that's fine, and once the majority of the world is on IPv6 then IPv4 addresses will be less scarce. I'd be more than happy with a CGNAT for IPv4, as long as I had direct connection for IPv6. The problem is chicken vs egg. Services can't afford to go IPv6 only while update is so low, and carriers have no incentives while services don't require it.
Posted by Dagger0 2 months ago
It doesn't matter that there's a bunch of v4-only devices and services out there. As an ISP, you still need to be doing v6 anyway, because deploying v6 is a necessary prerequisite to getting rid of those v4 devices.

Also, it's not that v6 isn't backwards compatible, but rather than v4 isn't forwards compatible. To talk to a v6 host, you HAVE to make changes to v4 hosts somehow.
Posted by TheEulerID 2 months ago
@Dagger0

Unfortunately ISPs all adopting IPV6 doesn't solve the problem as for some considerable time they will have to dual-stack for their customers. Indeed, dual-stack is the preferred migration technique.

As for IPV6 talking to IPV4 hosts, then that can be done with some translation services. None are perfect, but there are ways without changing legacy software.

Of course IPV4 can't be forwards compatible. When it was devised nobody had any idea what any future standard would look like.
Posted by Dagger0 2 months ago
It is, however, a prerequisite for solving the problem, meaning it's something you have to do first before you can solve the problem.

v6 talking to v4 hosts is certainly possible -- and you can do it right now, there's software that implements it and it's being used at scale by some large ISPs. You can't exactly argue that v6's designers made a mistake here when v6 already does what you're suggesting.
Posted by CecilWard 2 months ago
It certainly is possible for IPv6-only users to acces IPv4-only servers on the internet. The ISP Andrews and Arnold (aa.net.uk) offers this as an optional service, using protocol conversion, and it's free. Makes you wonder if this wouldn't be a much better option than scary CGNAT. I wouldn't go anywhere near a CGNAT ISP - that would instantly rule them out of consideration.
Posted by CarlThomas 2 months ago
Unless there's a one to one mapping between IPv4 and IPv6 addresses, Cecil, that still needs CGNAT.

If Hyperoptic had the addresses to do that they wouldn't be doing this.
Posted by bsdnazz 2 months ago
The actual compatibility problem is not backward compatibility but forwards compatibility.

IPv6 can map back to IPv4 addresses quite easily. The entire IPv4 address space (32 bits) can easily fit within an IPv6 subnet and IPv6 to IPv4 NAT style gateways could operate.

This means an IPv6 client could access an IPv4 service.

However, IPv4 cannot map to IPv6. Obviously within the IPv4 protocol it's not possible to specify all the IPv6 addresses.

This means an IPv4 client cannot access an IPv6 service.
Posted by chrysalis about 1 month ago
Not too surprising, they a fairly new isp, and as a result will not have had a chance to get excessive ip's when ipv4 was much more available.
But if they doing this without having ipv6 already on dual stack then it shameful.
You must be logged in to post comments. Click here to login.