Let’s talk IPv6, web hosting, and the future of networking

To say that talking about IPv6 is a bit like hitting a hornet’s nest with a stick is a bit of an understatement. It’s probably more like trying to pet a raccoon while yodeling in an avalanche zone. As a technologist running a tech company in 2022, it’s a topic where the responses shock, horrify, and astound me. There are several camps of people on the topic and I’m not going to really get into it because that’s not what this is about. It’s probably pretty obvious where I stand on the subject, but I wasn’t always there. Until rather recently, I had this stance of why use IPv6? We have plenty of IPv4 and everything nicely works for me. But then I was working on the  roadmap for NodeSpace and I realized, if we don’t take IPv6 seriously, we’re going to be left in the dust. Our operational costs have gone up significantly since I started this business over a decade ago, mostly within the last 3 years with some vendor consolidation in the web hosting space. Then, even more recently, ARIN has been becoming super-strict on IPv4 usage. And we don’t blame them.

IPv4 allocations are increasingly more expensive

This is supply and demand. It drives our economies. The gist of it is if you have a lot of demand for something but very little quantity, then the price is going to be expensive. This is where we’re at with IPv4. We have parts of the world trying to get connected, but IPv4 address space is mostly consolidated to large providers. These providers have so much extra address space, they lease it to others. This leasing adds costs. For example, these are arbitrary numbers, a /24 might have been assigned for $1,000/year directly from a RIR. But since the RIRs have none, it has to be leased by a private party. This private party might charge $7,000/year because they have a lot of people asking for IPs. And the next year, they might increase the price to $15,000 because of higher demand.

This is what we’re seeing. For example, we run a budget brand that uses a major global data center provider. For a long time, this provider only charged a 1-time fee for IPv4 addresses. Need a block of 8 IPs? No problem! Pay $30 once and you can use the block. Now, they’re charging by the IP and monthly because of their costs. Our budget brand just recently introduced IPv4 surcharges on primary IPs because the costs to get more have gone up.

Your home ISP might have to do the same. As they try to bring more subscribers online, their pool of IPs diminishes and to get more, they have to pay more. These costs get passed to everyone. So when your rate goes up, it’s partially IPv4 to blame.

When a customer asks us for more IPv4 space, it kicks off a lot of paperwork. We have to document why these additional IPs are needed and you have to explain them. This way, we can get more IPs from our network providers. Without this justification, they look at our usage and say “This customer has 20 IPs but it looks like they could just run with 1. Use those IPs. Request denied!”. This makes it difficult for us to be able to expand our services.

So following the law of Supply and Demand, when it comes to IPv6, we can offer services a lot cheaper because supply is high. Even if demand was insane, there is so much IPv6 space, we can just pull more from our allocation. High supply, high demand = still low price.

IPv6 is cheaper

Smaller and new ISPs are deploying IPv6-only networks because that’s all they have access to. They sign up for their AS from ARIN, and if they qualify as what ARIN refers to as a “3X-Small” organization, IPv6 fees are current waived through December 2026. If someone offered you free service you’d take it. And many of these services are doing just that.

Additionally, because IPv6 requires less overhead (you want a /64? Take it!), it’s cheaper to manage. ARIN doesn’t require IPv6 justification. Every server gets a /64! Easy! Service providers don’t have to do anything other than route the IPs and that’s it.

“If it’s not broke, don’t fix it”

This is something that I think a lot of people think IPv6 advocates are saying when it’s not. If you have an IPv4 network that is running fine, there’s no reason to shove IPv6 into it. No one is saying that you need to rip out IPv4 and replace it with IPv6. But if you find yourself utilizing a lot of IPv4 addresses and your bills keep going up, you may want to investigate if you could begin to phase in IPv6. Monitor your network – if you find more IPv6 connections than IPv4, you could scale down your IPv4 usage (and save money!). But believe me, no one is going to force you to rip out IPv4.

IPv6 at NodeSpace

Like I’ve mentioned, IPv6 has been on our 2022-2023 roadmap. We’ve done great work implementing and rolling it out. Has our deployment been smooth or perfect? Absolutely not. I’m looking at our internal issue tracker. I’ve counted at least 4 issues we have open regarding our IPv6 deployment. We’re working on fixing them! However, I do consider our deployment successful even if we are about 85-95% there. My goal is that we reach 100% IPv6 by Q4 2023. Many of our competitors have said that IPv6 is not a priority for them. Or because there’s no demand, it’s not worth implementing.

You know when it’s going to be a priority for them? When the cost of IPv4 addresses is no longer chipping away at their bottom line, but rather it’s blasting massive holes into it. This is why we started rolling out IPv6 to shared hosting and enabling it for resellers. There will be a day where we will shut off our IPv4 network. It’s not going to be anytime soon, but I see that ship starting to poke over the horizon.

Networking is also not a “demand driven” thing. Customers don’t drive that demand. Most end users don’t know or care about the technology that serves them up their favorite websites or apps, just as long as they can get to it. The “demand” driving IPv4 and IPv6 are other network engineers and the businesses they work for saying “our costs are going up” and then pointing at the IPv4 bill.

Why we advocate IPv6 use

First, we are a business and a business needs to make money. We identified we could provide more services to customers cheaper if we eliminated IPv4. As the business owner, I’ve sat down with account managers at data centers all over the country for quotes where we would love to extend services. But when 90% of the quote price is IPv4 networking, I flat out reject the quote unless we have the demand to immediately put the location into use.

IPv6 changes that. We could expand into more locations if we didn’t need to also provide IPv4 connectivity. IPv6 connectivity is so inexpensive for our partner data centers to provide, they don’t even invoice it. Or they discount the bill. We can pass these costs savings onto our customers. 2022 has been an expensive year and we certainly don’t want to pay more.

The more IPv4 lease allocations we can hand back, the cheaper our over all expenses become. The less we spend on IPv4, the more we can spend on better network gear and more data center locations. That, I feel, is a win-win.

IPv6 is here, it is the future, you need to learn it

And this is where I ruffle feathers. I get it. You learned networking with IPv4 and you’re comfortable with it. You just look at an address like 2345:0425:2CA1:0000:0000:0567:5673:23b5 and it’s overwhelming. IPv4 is easy and you’re knowledgeable on it. But do you know why you’re knowledgeable on it? Because you learned it! When you learn something, it becomes easier. I promise you, just like you learned IPv4, you can learn IPv6. And you’ll find out IPv6 is actually better than IPv4 in many ways. Did you know that there is no network or broadcast addresses? Yes, every IP is usable!

Another common thing I hear is Why is the address space so big? I don’t need 18 quintillion IPs! Well, if you learned IPv6, you would realize you actually do. I used to say this myself! But then, I realized something that was never really taught or expanded on and I’m going to quickly explain it. Maybe this is your “Ah ha!” moment. The short of it is a /64 prefix is the bare minimum for SLAAC to work (SLAAC is Stateless Autoconfiguration – basically, a better version of DHCP) and that’s because IPv6 can use a method to automatically generate an IP address from a MAC address. A MAC address is 48 bits. So the device will pad the address with some extra bits to get to 64. Much like in IPv4 slash notation, /64 in IPv6 means the first 64 bits are the network prefix. 64 network bits added to 64 host bits equals 128 bits. How many bits is an IPv6 address? 128. This was what finally made me realize why IPv6 space needs to be so big.

There are a lot of manufacturers of NICs so there are a lot of OUIs. And lets not forget about virtual NICs that have a virtual MAC address. There can be trillions of possible MAC addresses between two MAC addresses. And since these MACs can be used to generate IPv6 addresses by the host, and IP addresses need to be unique, you have to account for any MAC address to be on that network.

Once I understood that, it made sense why I needed a /64. Now I know on a dedicated server hosting websites, that also seems like overkill. But here’s the thing, IPv6 address space is so big, it’s practically unlimited. IPv4 taught us to conserve so when we start working with IPv6, we carry over those habits. And this is the rejuvenating bit that actually re-sparked my love for networking – we don’t have to care anymore! We don’t need to do complex math to figure out how many hosts will be on a network or potentially be on a network and do some VLSM math. We just say “here’s a /64!”

We’re conditioned to conserve all the time: electricity is expensive so in the winter time, we close our windows and we check our insulation to make sure heat stays in our home so our heaters don’t run. We shut off water to make sure we don’t waste it as only so much is actually drinkable. So in tech, we conserve our disk space so we don’t run out. We conserve our RAM usage so we can run other programs. But in IPv6, conservation doesn’t matter. If your computer was able to take an IPv6 address, use it for 1 second and then discard it and never use it again, it would take 584,542 years to deplete 1 IPv6 /64. You could use 10 IPs per second, and your children’s children’s children’s children’s children’s children would still have plenty of IPv6 space left. You have to train yourself to stop conserving in IPv6 world.

“I can’t use IPv6 because X”

This is just an excuse. “I can’t learn how to program because I’m too old”, “I can’t learn how to make a website because I only have an iPad”, “I can’t make videos for others because I only have my phone.” These are all excuses. Excuses are for unmotivated people. If we all made excuses, we’d still be sitting in caves drawing pictures on the walls and grunting. Sure, some services out there are only IPv4 – Twitter and Github are a couple I know off the top of my head. But these services are going to have to implement IPv6 sooner or later because of costs.

Final thoughts

I’m calling this section final thoughts although I know I definitely will not have any final thoughts on the subject. I suggest that anyone who has a negative attitude towards IPv6 approach it with an open mind. A must read is IPv6 Essentials 3rd Edition by Silvia Hagen, published by O’Reilly Media. ISBN: 978-1-449-31921-2. Pick this book up in physical or virtual format. Read it cover to cover.

Then, get your hands on some IPv6 address space or a server with IPv6 address space. Play around with it. Learn with it. Most importantly, keep an open mind and remember that IPv6 is not IPv4.

Happy learning!

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.