IP addresses

Will it be possibe sometime in the near future to run out of IP addresses? Are there any plans in the works to address this issue?

Yes and yes. The last blocks of IP addresses have already been assigned, so we’re running out now. A new scheme with a longer address number exists, and is being implemented, though not as quickly as some people would like. As a stop-gap, there have been an increase in networks that use a single gateway to attach several computers to the internet, and thus conserve IP addresses.

The new protocol is IPv6,

We’ve already run out of IPv4 addresses, which is the version of the Internet most of the world is still using: The last blocks have been assigned, which means someone already owns every single IPv4 address that can be used on the Internet. (Some blocks of addresses are not usable on the Internet, as they have been reserved for special purposes such as giving IP addresses to machines on local networks.)

IPv6 is the new standard, which is in use already but not to a huge extent. (IPv5 was an experiment that never made it into the wild.) The process of moving everything to IPv6 is either already in progress or it’s never going to happen, depending on who you listen to.

I hate dashing into a thread to do this, but in before obligatory xkcd reference.

No and yes.

Derleth is correct that all ipv4 blocks have been assigned, but that doesn’t mean we’ve run out. It just means they’re a bit more valuable. Large swathes of addresses are unused, hoarded. Addresses get delegated, reassigned and loaned out all the time.

We’re not even remotely close to running out in the sense that they’re all in use and there’s no more room on the internet.

This is true. I have a beta version of Firefox that lets me see the ip address of the computer of everybody logged onto here right now and it’s plain to see that we’re not using anywhere close to all of them.

First, how would that even work? Is it getting data from the SDMB servers? Have I misunderstood you?

Second, assuming it is what you say, that’s a highly biased sample you’ve got there and I would not draw huge conclusions from it.

I read somewhere that the MAC address (which is unique for every net card) is one of the few instances where the internet founding fathers actually guessed right. Even assuming exponential growth the current scheme, which allows for over a quarter quadrillion unique assignments (2[sup]48[/sup] or 281,474,976,710,656), should be enough to get us thru to 2100 A.D.

Which is really irrelevant anyway, because MAC addresses only matter on your local Ethernet segment.

I feel I should know but it isn’t coming to me. Why is the year 1998 relevant in this strip?

In the strip they’re grateful for the limit on the number of possible adresses in IPv6, which became a standard in 1998. There are “only” about 3 * 10^38 addresses; it’s on the same order as the volume of Earth in cubic microns.

The requirement is different. MAC addresses have to (by design) be unique across the span of time, so no 2 Ethernet devices can ever have the same MAC address.

IP addresses only have to be simultaneously unique. 50 devices can have the same IP as long as they don’t all have it at the same time.

The problem with IPV4 address exhaustion is one of greed and mismanagement, not one of running out of addresses. The 32 bit IP address has to traverse every link from here to the web server that happens to be storing this load of bollocks. That uses valuable resources. The Ethernet MAC doesn’t work quite the same way.

Only 2 ethernet devices on the same subnet HAVE to have different addresses. DECnet in fact would rewrite the MAC address as something different reflecting the DECnet address, IIRC. The celver plan was simply to avoid having to have a way to generate guaranteed non-duplicate numbers. As pointed out, it was clever and very functional.

In my time, “Ethernet” has always been 48 bits. Always. You can change the MAC if you want, and for security reasons you might want - but can’t change it to less than 48 bits. You can’t change it to a 3 bit identifier, for instance, even if you do change the value to 101. That only makes it 101 with 45 zeros in front.

Since IEEE guarantees that the MAC address is unique and has for some time now, I wouldn’t be too eager to change my MAC. Presumably we’ll have another system in place before 2100.

One thing that I would take exception to is any implication that “more is better”. Bigger identifiers make bloated protocols make poor performance. Size it right - not too big, not too small. If I had to make a judgement i would say the IEEE 802 48 bit MAC standard was too big.