The internet has run out of IP addresses and what happens after that is anyone's guess

reply to post by TLomon


What do you think?

ipv666.com
Ipv666.com is a domain controlled by two name servers at ipv666.com themselves. Both are on different IP networks. The primary name server is ns0.ipv666.com. Incoming mail for ipv666.com is handled by one mail server at ipv666.com. ipv666.com has one IP number (24.121.227.32).

Air-control-ac.com point to the same IP and also shares name servers. Mail.air-control-ac.com and www.netcommaz.com point to the same IP. Ipv666.com use this as a mail server. Netcommaz.com, vwadventures.com and ipv11.com share name servers with this domain. Net.ipv666.com, ns0.ipv666.com and ns1.ipv666.com are subdomains to this hostname.

www.robtex.com...

Originally posted by SpaDe_
reply to post by phishybongwaters

 

That is my point. There can never ever mathematically be an ending number of any given IP addresses. Just because this current version being based on 32 bit is running out there is always the next set. Next after this will be 256 bit and so on and so forth. It is never ending.

1. Of course in one sense you are abolutely right, but the issue remains, the same as Y2K. Y2K was an issue because programmers only used two digits to store the year. The reason was not laziness, but memory space on early computers where every byte was precious. Programmers relied on the two digits and performed their mathematical calculations accordingly. No one in 1970 believed their programs would still be running in 2000, so it was not deemed a problem. Except that it was and when 2000 rolled around, lots of people were born in negative years.

One of my programs, an accounts receivable/payable program, suffered from this. I had to go back and "patch" the program by adding a couple of lines of code that circumvented the issue. I did and everything worked fine. That's what a lot of other prgrammers did, too. It was also a great excuse to upgrade. One company I was associated with, WLN, the Western Library Network, folded because their main "Grinder" program was so full of Y2K issues that they felt they did not have the money to fix it, so they "merged" (were absorbed by) OCLC. They never said, "Y2K drove us out of business." Instead they made noises about consolidation and better use of funds, blah, blah, but Y2K was what did them in.

Y2K WAS a problem, but people like me fixed the issues and insulated you from them so that you could then claim it wasn't an issue. There's plenty of ignorance to go around here.

2. IPv4 is exactly the same problem. A "header" is a fixed length piece of data that is attached to ALL the packets sent out on the Internet--ALL of them. Version 4 addresses are composed of four (4) bytes, and ONLY 4 bytes. Each byte can hold one of 256 possible numbers, so your range goes from 000.000.000.000 to 256.256.256.256. That's all you've got. The header contains the destination IP and the origin IP so there is more than one place in the header where this happens.

This is not pure math here. You can't just cavalierly say. "Well, just add a fifth byte and you've got all the address space you need" because the header of the packet is a fixed length with every byte accounted for and every machine on the Internet poised to look for 4 bytes, not 5. This is practical math, and as I hope you can see, we have a practical problem to solve.

IPv6 was designed to solve this issue, but it must be implemented. That means "touching" every single router on the Internet. Just be glad we're not talking MAC addresses here, which are burned (hard-wired) into every hardware device on the Internet. The solution will be in software to re-program routers to handle the larger address space required. The good news, as has been stated, is that the version 4 addresses can be treated as a subset of version 6 addresses, so no one has to change anything locally (save those pesky routers). It will be an issue with big corporations and ISPs, but it will not be a problem for the public.

3. Just like the Y2K problem, no one ever thought we'd run out of IP addresses. That was before fax machines, copy machines, printers, and every other electronic thing started having IP addresses. In the early days the "authorities" were very cavalier themselves about sending out addresses. In the eighties I had ten outlets to serve, so I asked for eleven Class "C" networks. A "Class C" network allows you 256 (well, no, it's 254 because you don;t use 0 or 256, but let's not quibble) addresses per network. So, for example, one network was 198.187.135.1-255. It's just that last byte that would change in the "198.187.135 network" I would assign one constant number to my router, such as "dot one" and use the rest for my computers.

The problem was, in some of my networks, I only had ten computers on them, but I used an entire Class C on them. That means I wasted 250 addresses which are sitting there (still) unused and unavailable,. I also had an entire Class C for my WAN to tie all the Class C networks together. I would guess that at first I had something like 100 devices and 2750 addresses, therefore 2650 of them were unused.

This is an INCREDIBLE waste of address space and it is prevalent today. Although with the fax, printer, telephone issue has grabbed many of those addresses today to the point that I began to sub-net (a way to divide a Class C into separate networks), the fact is that today that combination of networks still is using only half its address space.

Are there ways around this? Sure. I used static addressing and permanently assigned IPs to devices. I could have used dynamic addressing where a computer assigned a temporary address to each device as needed. The problem was that I used the IP as an identifier. It would have been extremely difficult, time-consuming, and expensive to switch procedures. I could also have used internal addressing where the addresses were not advertised on the Internet. This would mean that for my now 1,000 device network I would use just a few dozen externally. When I originally designed the network some of those options were not open to me. Today it would be considered unprofessional not to use them.

Could the authroities just say, "Give them back!" No. They are mine. Besides, technically it would be extremeley difficult. Those networks are sub-netted every which way these days. It would be a technical nightmaree. Besides, who says I'm not using them and I still need room for gorwth.

The problem here is that anyone NEW asking for NEW addresses will not be able to get version 4 addresses any more. They are all used up. THAT'S why this is an issue. Addresses have become valuable. My guess is that some enterprising souls will come to people like me and say, "Look, if you consolidate and do DHCP instead of static addressing that will free up some addresses we can jointly sell to the new guys." All we need is an enterprising spirit

So the bottom line point here is that decisions have consequences and sometimes you don't know what they are until you face them years later. And although IPv4 is "Not a problem" as far as you, the end user and customer, that doesn't mean there aren't lots of people working very hard to ensure it doesn't become a problem for you so that you, once again, can feel no effects and conclude that it was never really a problem anyway.

Well, it was.

Originally posted by LanMan54
reply to post by TLomon

 

What do you think?

ipv666.com
Ipv666.com is a domain controlled by two name servers at ipv666.com themselves. Both are on different IP networks. The primary name server is ns0.ipv666.com. Incoming mail for ipv666.com is handled by one mail server at ipv666.com. ipv666.com has one IP number (24.121.227.32).

Air-control-ac.com point to the same IP and also shares name servers. Mail.air-control-ac.com and www.netcommaz.com point to the same IP. Ipv666.com use this as a mail server. Netcommaz.com, vwadventures.com and ipv11.com share name servers with this domain. Net.ipv666.com, ns0.ipv666.com and ns1.ipv666.com are subdomains to this hostname.

www.robtex.com...

I am not sure what I should be responding to. You are detailing how DNS works. One IP address can host an infinite amount of names, if someone is willing to register (and pay associated fees) for it. www.a.com, b.com, etc. A DNS entry can also go directly to a subdirectory. So, 127.0.0.1/stuff can map to www.mystuff.com. Anyway, I need to get out of here. I will respond more tomorrow.

reply to post by TLomon


If i hear you right you are correct. I am a network administrator as a job, its impossible to run out of ips. IPv6 is pretty much unlimited, you'd have to try incredibly hard to fill one of those up.

This is just ill informed people thinking they have something, when they just don't get it at all.

reply to post by doom27


First I want to say thanks to schuyler for that excellent description of the Y2K problem. Many people thought at the time, and most people still think, it was just a lot of hysterical... uhhh... bullfeet. I worked on that problem myself. COBOL programmers were being hired like crazy; you could pretty much write your own ticket.

But I'd like to hit the concept a little harder of how many IP addresses 128 bits really represent. It's one thing to say, "oh that's 3.4 times 10 to the 38th," or whatever. But consider this: Google alone asked ARIN (the American Registry for Internet Numbers) for some 7.9 times 10 to the 28th (or 2 to the 96th) IPv6 numbers--and got them. Don't bother with the math: that's 79 BILLION BILLION BILLION addresses. And that's just one company. (I mean, it's Google, of course. But still....)

And there's plenty more where that came from.

So, back at work, and here are some additional replies. First off, schuyler - wonderful post. Star for you. I was in the trenches during Y2K for about 3 years before the "big event". It was definitely an interesting time. What really amazed me was the type of equipment that was seriously affected by the date change and you wouldn't even expect it to even know the date. (For example, a sewage pump - it had date stamps in its log files, and when the date change occurred, the pump switched into reverse - not good.) Shoot, one company I worked for had a system crash every year that the previous developer would go in change a hardcoded date - it was effectively no-year compliant. It amazes me that some people found this acceptable.

IPv6 will practically never run out - but we thought the same thing of IPv4. As more devices are being added, who knows what the future will entail - but at that time we should be on IPv8, or something even higher.

Regarding IPv4, schuyler is right. There is a ton of wastage. The one that blows my mind away is why we have a reserved Class A for loopback testing. 127.0.0.1. 127.x.x.x (if you look at the binary) would be 0111-1111.0000-0000.0000-0000.0000-0000 to 0111-1111.1111-1111.1111-1111.1111-1111. (Thus making it a Class A if it was a fully utilized IP address). This equates to 16,777,216 (ok, 214 - need your network and broadcast) addresses for loopback testing. How in the heck could you ever use that many? At least the 10.x.x.x one makes sense - it is commonly used for intranetworks within a company. The company would use NAT (Network Address Translation). This would then utilize a public IP, such as 105.1.23.158, and convert it to an internal network formed entirely of 10.x.x.x devices. NAT is one of the major reasons why we have not converted over to IPv6 sooner.

Most of the world is ready to go for IPv6. So, who/what is the hold up? It is quite simple - the United States. When IPs were originally allocated, the US took the lions share. Third world countries often times shared an IP range. With the population growth and the widespread usage of technology, the countries now need a lot more - they are chomping at the bit to do so. Because the US had so many to begin with, we have been holding off a conversion that most of the world is already to go forward with. However, a lot of companies have already done the work - mine is ready to go at the "flip of a switch". We still have some legacy devices we are trying to get updated - and they will - but IPv4 will route over IPv6 without problems.

Now, back to IP wastage. Let's look at a small IP block.

192.168.0.x. This would be 256 addresses (0-255). However, you need to use 0 for network, and 255 for broadcast, so effectively, this is 254 addresses. Originally, subnetting would let you break it down. So, let's use 64 blocks.
192.168.0.0-63
192.168.0.64-127
192.168.0.128-191
192.168.0.192-255
Each of these needs a network and broadcast, so we just chopped down a single IP block into 4 subnets, lowering the 254 addresses to 248. However, this now allows us to use the IPs for 4 different networks. Each network would have 62 addresses. This could be assigned to 4 different customers, etc. But what if a customer needs only 4 IPs? They just wasted 58 IPs that could be assigned to someone else.

Then came variable length subnetting. This allowed us to break down subnets into smaller groups.

192.168.0.0-3 - this is a point to point connection (server A to server B). It uses 4 IPs but with the two required, you only have 2 usable ones.

192.168.0.8-15 - notice the jump. To issue a set of 8 IPs (6 usable), you have to start at the appropriate block size. We couldn't start at .4, because that would force an IP range that overlapped the first subnet, or limit it to 4 as well.

I could go on and on. It is nice to see some peers who realize just how misinformed the writer of that article was. I wasn't sure how many people would agree with me on that subject.Anyway, I hope I explained this well enough for non-engineer types. If you ahve any questions, feel free to post here or send a PM.

Lol so much rubish,...most ppl who have windows 7 have IPv6 running allready if you look at your network connections

4 Digits ? sack the reporter it has 4 Octets