Did IPV4 run out?

[aimeeandbeatles]

According to a counter I saw on a website it said "Exhausted." Did they run out? Google isnt helping very much.

 

[Maniacal]

nevermind

 

[Speedo]

My understanding is that all of the never-before-used blocks were handed out in the last few months. They're still recycling addresses that had been previously used.

 

[mdwh]

They're handed out in a hierarchy. All of the blocks were allocated by the top level (Internet Assigned Numbers Authority) in February. That doesn't mean they're used up yet, as there are still some available at the lower levels. But the rest are expected to run out quickly. See http://en.wikipedia.org/wiki/IPv4_address_exhaustion .

 

[taper]

They've started setting up markets for IPV4 address. It's sad but not surprising that IT managers worldwide have pretty much ignored IPV6.

 

[Turner]

Well, it's not the threat that Y2K was.....

 

[mdwh]

On that note, if the transition to ipv6 had or does somehow manage to happen smoothly because the people concerned do the job right, we'll then just hear people moaning "Oh, what a load of scaremongering nonsense that talk about ipv4 running out was. We didn't have any problems after all!"

 

[Turner]

I don't think so.... Everyone pretty much acknowledges that Y2K was a serious threat and that a lot of effort went into correcting it.

Of course, this is the equivalent of a brown-out, not the black-out Y2K could have been.

 

[deanej]

I've never understood why Y2K was an issue in the first place. Why does it matter if a clock thinks it's 1900 or 2000? And if it is, it can easily be worked around:

if(year < start_date) {
    year += 100;
}

Of course this only prolongs the issue, but a century is more than enough time to fix the system.

 

[mdwh]

I don't think so.... Everyone pretty much acknowledges that Y2K was a serious threat and that a lot of effort went into correcting it.

Of course, this is the equivalent of a brown-out, not the black-out Y2K could have been.

I agree with you that it was a threat - but it's not uncommon for me to see people assuming it was just an overblown myth, on the basis that "nothing" happened.

 

[mdwh]

I've never understood why Y2K was an issue in the first place. Why does it matter if a clock thinks it's 1900 or 2000?

Well, admittedly it wasn't clear to me why say a piece of electronics might fail. But in general, plenty of chaos could be caused by getting the dates wrong - e.g., date of births, banking information.

And if it is, it can easily be worked around:

if(year < start_date) {
    year += 100;
}

Of course this only prolongs the issue, but a century is more than enough time to fix the system.

Sure, but the not so easy bit is checking every piece of code in millions of systems to see what needs fixing.

Plus I think Y2K and IPV4 show us that no matter how long we have to fix the problem, it gets left to the last minute...

 

[Speedo]

I've never understood why Y2K was an issue in the first place. Why does it matter if a clock thinks it's 1900 or 2000? And if it is, it can easily be worked around:

if(year < start_date) {
    year += 100;
}

Of course this only prolongs the issue, but a century is more than enough time to fix the system.

The systems that had the problem only understood a 2 digit year. Adding 100 would've had no effect.

Why is it an issue? A thousand reasons. Anything that was only valid within a certain date range wouldn't work. Credit/debit card processing. Timing or monitoring equipment.

If you don't think it was an issue, take a look at the wiki article and the dozen or so examples it lists of (minor) problems that did happen. The world didn't spend several hundred billion dollars fixing Y2K issues just for fun.

 

[deanej]

Can't they train the system so that everything else rolls over too?

 

[aimeeandbeatles]

Actually, I read that when Y2K rolled around there were only minor issues (such as websites showing 20100 for example).

 

[Maniacal]

Both Y2K and the IP thing seem needlessly complicated issues they should permanently fix.

 

[Speedo]

Can't they train the system so that everything else rolls over too?

...Train the system? I assume you mean program it so that everything rolls over? I guess it might work in some cases, but a lot of systems needed to be able to work in the current century while still being able to handle records from the last century. So they need to recognize both 19XX and 2XXX years.

Actually, I read that when Y2K rolled around there were only minor issues (such as websites showing 20100 for example).

Yeah, because companies and governments around the world spent something like 300 billion dollars doing fixes.

Both Y2K and the IP thing seem needlessly complicated issues they should permanently fix.

The nature of the way computers handle data (at least, our current computers) means that a permanent fix is essentially impossible. The fixes we have just push the next problem far out into the future.

Using a 4 digit year means we have 7989 years to come up with something better before our grandchildren to the nth power have to worry about "Y10k" bugs, while IPv6 supports 2^128 address as opposed to the "mere" 2^32 addresses in IPv4.

 

[Petek]

I think that the next widespread computer problem is not that far off: The year 2038 problem.

 

[deanej]

Depending on 64-bit adoption this might wind down on its own. But the solution is simpler: either change to a 64-bit program (where the compiler will fix it itself), use an unsigned int, or use a long long int (which is essentially an abstraction of a 64-bit int done by doing math in 32-bit parts).

 

[civ_king]

Depending on 64-bit adoption this might wind down on its own. But the solution is simpler: either change to a 64-bit program (where the compiler will fix it itself), use an unsigned int, or use a long long int (which is essentially an abstraction of a 64-bit int done by doing math in 32-bit parts).

Embedded systems must be replaced and can't be fixed software wise