End of the road in sight for IPv4?

[x88x]

I found this interesting article (http://arstechnica.com/web/news/2010/10/well-of-remaining-ipv4-address-blocks-quickly-running-dry.ars) today on the current status of world public IPv4 addresses...turns out it's a lot worse than I thought it was. Long story short, at the current consumption rate we have maybe 2 years until some regions start running out. :eek: Thankfully, this shouldn't affect any normal consumers, so most of us can just sit back and watch the show. It'll be interesting to see what steps different regions take to solve the problem over then next few years. Do we transition to IPv6 or do we start making early registers sell back unused IP ranges (multiple A-block holders, I'm looking at you -_^ ).

[slaveofconvention]

I was reading something recently about the expected problems with IPv6 - despite the huge range available there's now concern about THAT running out. Newer internet connections are expected to effectively be backbone connections so every device on your local network will have it's own dedicated IP address as opposed to the current private IP within your own network. As more and more devices become internet connected (I personally already have about 2 dozen internet connected devices in the house) there may end up being issues with even this huge quantity of addresses...

[mDust]

I was reading something recently about the expected problems with IPv6 - despite the huge range available there's now concern about THAT running out. Newer internet connections are expected to effectively be backbone connections so every device on your local network will have it's own dedicated IP address as opposed to the current private IP within your own network. As more and more devices become internet connected (I personally already have about 2 dozen internet connected devices in the house) there may end up being issues with even this huge quantity of addresses...

IPv6 provides 2^128 addresses...or assuming the human population is 7 billion, that's over 48,611,766,702,991,209,066,196,372,490 addresses per person. Even considering corporate and backbone needs, I think that's plenty.

I think we should reclaim unused addresses while we finish upgrading POS equipment that still doesn't support IPv6.

[Drum Thumper]

Due to the way subnetting and IPv4 works, reclaiming is impossible.

[x88x]

Due to the way subnetting and IPv4 works, reclaiming is impossible.

Could you elaborate on that? I don't claim to be an expert on the topic, but I know pretty much how it works and if done properly I don't foresee a problem.

Take MIT for example. I figured out a while ago what all they had, but I forget what it was now.. I know they have at least the entire 18 A-block (or /8 block, whichever terminology you prefer). That's 18.0.0.0-18.255.255.255 for those who don't know the terminology. I'm pretty sure I remember finding out that they also have at least one more A block and a few B blocks as well, but let's pretend they only have this one. That's a total of 16,777,216 addresses (ok, 'only' 16,645,630 useable, but still, that's a LOT). Now, I know MIT does a lot of stuff, but I rather doubt that they have any real need for over 16.5 million public IP addresses. Depending on how they have assigned their addresses over time, it is entirely possible that they have a B-block (ex, 18.5.0.0-18.5.255.255) or C-block (ex, 18.5.5.0-18.5.5.255) completely empty. This block could then be sold back to IANA and sold to somewhere that actually needs those addresses.

Granted, I don't have personal knowledge of the internet footprint of MIT, but I do know the internet footprint of the college I used to attend full-time. They are a small private college with probably about 3,000 students living on campus, maybe 1,000 commuter students, and maybe 1,000 faculty and staff (I think it's actually lower than that). Add to that, iirc, about 200 computers in various labs around campus. So, assuming everyone has one internet-connected device, that's a minimum required capacity of about 4,200 addresses. Just for the sake of argument, let's say that maybe every student living on campus has 10 internet-connected devices (completely unrealistic, but let's say that's a max), and each commuter student and faculty/staff has 2. That brings a likely max required addresses to 34,200. This college leases an entire B-block from HP (I won't say which one, for privacy's sake...and since last I was there they had pretty crappy security...). That's a total of 65,536 public IP addresses, 65,022 of which are useable (dropping the .0 and .255 addresses). Granted, future growth should be allowed for, but still, there is no way that this college needs that many public IP addresses. In fact, I would argue that they don't need more than a few public IP addresses. Yes, sometimes it's convenient to have every computer I own have a different public IP address, but there's no real reason for every device on the network to have a public IP address. Seriously, network admins, manage your network properly and there's no reason why you need more than a few (often no more than 1), no matter how large your network is.

[Diamon]

Maybe what x88x says in combination with ISP NAT's (which would make it a hell to track IP's ^^) will keep the IP's from running out for at least a few more years. Unless someone comes up with a brilliant idea we'll have to switch to IPv6 sooner or later though.

[x88x]

I don't really like the idea of ISP NATing (even if it would make tracking harder ;) ), I like being able to get to access my SSH server from anywhere without special software.

[Diamon]

I don't really like the idea of ISP NATing (even if it would make tracking harder ;) ), I like being able to get to access my SSH server from anywhere without special software.

Me neither. But it's one of very few good ideas to make the IPv4 addresses last longer.

[mDust]

I think it's funny how over 20 years ago this problem was foreseen and IPv6 was offered up as a solution. Ten years after that they released the IPv6 standard and said 'Problem solved! Let's begin the switch." And now here we are today, another 10 years later, and we're still worrying about it. Could the organizations in charge procrastinate any longer? I hear that a lot of the US backbone infrastructure is outdated and running extremely close to maximum capacity, so what are they waiting for?

EDIT TO AVOID DOUBLE POST:
Apparently it is possible for organizations to return large unused blocks of addresses. Not only is it possible, but commended by ARIN. Sixteen million addresses extended the impending disaster only 1 month though...so everyone else holding on to them for no reason needs to do the same.
https://www.arin.net/announcements/2010/20101020.html

[dr.walrus]

In fact, I would argue that they don't need more than a few public IP addresses. Yes, sometimes it's convenient to have every computer I own have a different public IP address, but there's no real reason for every device on the network to have a public IP address. Seriously, network admins, manage your network properly and there's no reason why you need more than a few (often no more than 1), no matter how large your network is.

This. Four computers on my home network, two xbox 360s and an FTP/HTTP server, one public IP address.

If nothing else, it's actually cheaper and more secure, because I only need to firewall the router, not all the other devices... And I'm unlikely to be using bandwidth intensive stuff on more than one device at a time.