Well, here we go again. Today, CNN is 'wondering' whether we are ready for the next big internet crunch: the run out of IPv4 addresses !!! The time clock is set at September 2011:
This growing problem has been compounded in recent years as the amount of hardware needing an internet connection has risen enormously, thanks largely to the popularity of smart phones, like Apple's iPhone and other mobile devices, and the rapid pace of technological development in countries like China and India.
Bull-crap !!! What is Campbell Brown's phone number again? No, I mean seriously? Which facts are supporting this sudden stress? It's breaking news !!!How about a little sanity check? The latest stats indicate about 1,8 billion internet users worldwide. There are about 3.7 billion available IPv4 public addresses. So, a potential of about two IPv4 public addresses per user. And, how many of us are actually connecting to the Internet with a public IP address? Raise your hand if you do! What about NATs allowing us to (potentially) demultiply each public IPv4 addresses by 63K? How many devices per user would that be?
Let's get real here. The article stresses that there is a solution (IPv6), but that people and companies are wrong to not anticipate the dramatic impact of not handling this upcoming crisis in time. I guess we are all living in lalaland !!!
Here is my bet about what is REALLY going to happen:
- When things are easy, people are lazy. People and companies would rather acquire new IP addresses than recycle existing ones.
- A potential of two public IPv4 addresses per user worldwide indicates that current allocation is sub-optimal and that a bit of management would help.
- When supply will run-out, people and companies will start recyling unused IP addresses. They will also consider moving devices who don't really need a public address behind a NAT - How dramatic is that?
- People and companies will start trading IPv4 addresses as long as it remains cheaper than putting money into new IPv6 equipement and upgrading applications - Let's create a derivative financial product to hedge costs !!! There is money to be made !!!
- What if trading becomes too expensive? Most routers can already publish both a public IPv4 and an IPv6 address. They are not part of the issue. Now, how hard is it to upgrade an IPv4-enabled application to IPv6? How much code re-writting and testing is required? Does one need to rewrite an IP stack? Aren't these ready? And, how many automatic updates do you install per year on your PC? Can't we use that venue? - Let's declare national emergency !!!
- The IPv4 to IPv6 transistion will go smoothly...
Oh dear..
ReplyDeletePlease tell me you have more than half a clue about IP addresses before ranting about CNN having no clue about IP addresses? It's blocks of addresses that are key, not individual ones.
IPs are split up into subnets. I'll take a private range by way of example:
192.168.0.0/24
that /24 marks out a subnet, it means the following usable IP addresses: 192.168.0.1 - 192.168.0.254
Can all be summarised to the outside world as 192.168.0.0/24.
That's critically important. It would be impossible to store the routing details for every single IP address in every border router on the internet.
An ISP might end up with a larger block like this:
192.168.0.0/16 (191.168.0.1 - 191.168.255.254) Which they sub-allocate to customers, blocks at a time. That way instead of needing to advertise the route to every /24 (all 255 of them in that /16), they could just advertise that single /16 to the world. Any border router needing to send data to any of the /24s would know that the /16 covers it.
So take a big ISP like Time Warner. They get allocated blocks of IP addresses. What if their customers aren't using all of them? Doesn't matter. They're "wasted"? Not exactly.. you just can't allocate them to anyone else without making the internet a huge slow ugly mess.
Summarisation of addresses and blocks are absolutely critical to the smooth running of the internet.
Every single packet of data at every single border router gets run against all the routing table. That takes processing power. The more entries there are in the routing table the more checks need to be done, the higher the latency.
What we're rather rapidly approaching is a lack of spare blocks off addresses to allocate.
Your ideas of fixing the problem are based not in anything approaching reality, but in something that will see internet become totally useless.
Taking a look at the problem from blocks of IP addresses or individual addresses does not make a difference, globally speaking... Constraints and solutions remain exactly the same!
ReplyDelete