IPocalypse- by Jose Robles

It’s been called the IPocalypse.  The catastrophic event where the world will run out of traditional IPv4 addresses once we exhaust all of the available unused blocks.  It’s a funny tongue in cheek term which is used to try and instill a sense of foreboding about the looming work and changes required to move onto the next version of the protocol known as IPv6.

So how did we get to this point where we feel that we’re scrambling to keep the internet alive?  To be clear, the internet is not going to shutdown on the day that IPv4 space runs out.  This is not like the Y2K scares from 10+ years ago even though it has been compared to that mess.  All will continue to route fine and things like Google, Twitter and Facebook will continue working.  In fact, it’s been discussed many times by experts that IPv4 and IPv6 will coexist for a great number of years.  Even decades!  But I get ahead of myself…

IPv4 was created back in the 1970s and had an increase of use in the 1980s.  At the time, four billion IP addresses were seen as being more than enough to handle our needs for connectivity between all of the computers/servers that were “online”.  As the popularity of the Internet exploded in the mid to late 90s, more and more devices required having their own public IP address in order to access online content.  Think back to the dial-up days.  Chances are your ISP was giving you a public IP address.  As services moved to broadband speeds and more of the population got Internet access, ISPs began to use even more of their public IPv4 blocks to get customers online.  It didn’t take long to realize we would eventually run out and in fact organizations were already working on the next version of IPv4 back in the early 90s.

After a “few” debates and discussions about how the next version would work, IPv6 was announced and released as a standard.  While IPv4 addresses are 32 bits in length and have about four billion addresses (4,294,967,296), IPv6 addresses are 128 bits in length and seems almost limitless in terms of the sheer number of addressess available (340,282,366,920,938,463,463,374,607,431,768,211,456).  There have been many metaphors used to try and relate to how large this number really is such as:

  • Every human being on the planet can be assigned trillions of IP addresses
  • You can assign an IP address to every single atom on the planet
  • If the earth were made entirely out of 1 cubic millimetre grains of sand, then you could give a unique IPv6 address to each grain in 300 million planets the size of the earth

My personal favourite and likely easiest one to show is this.  Imagine that all of the IPv4 space fits inside the space of a golf ball.  That’s IPv4.  IPv6 is the size of the solar system!  It’s still hard for us humans to envision something so expansive but suffice to say, we have a lot of IP addressing with IPv6 but we should still be clear that there are limits to it as well since it is a finite number.  Chances are though that we won’t see or come across these limiations within our lifetime.

As of January 2011, there are seven /8s (/8 = 32 million addresses) remaining in the global pool which is maintained by IANA.  Once the pool reaches five, the reamining /8s are then handed out to the five RIRs (ARIN, AfriNIC, APNIC, LACNIC, RIPE) and then there are no more new blocks to hand out…from IANA.  The estimated date for this event to occur is February 2011.  Once the RIRs have received the last of the /8s, they in turn have either months or years before they exhaust their allotment.  Areas like APNIC will like go through their remaining IPv4 blocks in months due to the sheer number of devices going online that require public IP space.  AfriNIC has been estimated to be able to last for years because they do not have as much demand for publically accessible devices.  It’s safe to say that North America will likely reach complete exhaustion in one to two years from now.  This is the main reason why we need to be active on IPv6 development and deployment right now versus waiting even further than what we already have as a global community.  What would happen if we didn’t implement anything?  Well, business would continue as normal but your customer base would have plateaued as you would not be able to handout new IP blocks to any new customers so you can’t generate any new revenue.

So how do we get there?  What do we do?  Well, there are various methods for implementing IPv6 and providing new connectivity for your customers but it all depends on the organization.  ISPs will likely want to implement at minimum a dual stack rollout where both IPv4 and IPv6 coexist together.  This would likely stay in place for years if not decades until everything is transitioned over to IPv6.  What this method gives you is the ability to continue to provide traditional IPv4 access to all websites and services while giving the customer the ability to reach new sites and services that are running IPv6.  Other methods include tunneling (IPv6 over IPv4) and proxying either IPv4 to IPv6 or IPv6 to IPv4.  These other methods are the ones that scare the industry as they usually involve more capital investment and greater education to get it done.  So would you end up having to spend billions of dollars to get things going?  Again it depends on the organization.  Consumer/residential ISPs may have to implement the NAT/proxy solution in order to continue providing a service to their customers.  Business related ISPs might be able to bypass that…maybe.  That being said, there are a handful of ISPs in Europe that have transparently (to the resi customer) and successfully provided IPv6 connectivity via their CPE/modem.  This way, a home user doesn’t have to worry whether their machines have IPv6 capability or not (they probably do by now).

The rollout of IPv6 on a provider’s network could take months or years to get it done right.  This is an opportunity to rollout assignments in a proper, manageable format versus the adhoc method that was sometimes used when scrambling to turn up new customers in new markets.  Proper aggregation of your blocks to your BGP peers is key here.  End users wishing to try out IPv6 can already do so today with the many tunnel brokers that exist which will provide you with a free IPv6 tunnel and your own IPv6 space to assign to your LAN.  This is a great way to get a feel for how things will eventually work.  Although it’s geeky, it does put a little smile on your face when you can get to something like ipv6.google.com or www.v6.facebook.com on your browser.

So what happens after everyone adopts IPv6 on their network?  Well, as I stated earlier, IPv4 isn’t going anywhere anytime soon.  There are simply way too many devices out there that either do not support IPv6 right now and need a software upgrade or will simply never support IPv6.  These devices will likely serve out their five year lifespan of most electronics and be replaced with new ones that do support the new protocol.  It is reasons like this that we need and will continue to support IPv4 for the coming decade…at least!  To sum things up, start learning about IPv6 right now.  There are many places online that give you very good tutorials for getting at least the fundamentals of it going.  Start auditing your network to see if there are hardware/software upgrades you will need to do in order to support IPv6.  Test, test, test.  And rollout based on whatever methods best suit your organization’s needs.  The next couple of years will definitely be fun for IT and Network admins.

Comments

  1. Great primer and overview of the IPocalypse Jose. I know from a network pov it’s more pressing, resolving this issue, but I’ve always stated that IPv4 isn’t really running out. How many A blocks are being managed by the military and other governmental institutions? I highly doubt they’re using every IP availed to them, but of course, they’ll never give them up.

    Nonetheless, you’re bang on. We’ve been talking about this since the 90s and no one that I’ve heard of has been really prepping for this in earnest.

  2. Great article Jose!
    I always wondered if IANA would somehow go back to the organizations that were handed out Class A addresses in the 1990’s and ask for them back. I bet there are hundreds of government and private organizations who are sitting on IPv4 address blocks that they will never need
    :-D

  3. Jose Robles says:

    Thanks for the kind words Jules/Karen. :)

    Yeah there’s been talk about the organizations that still have /8s all to themselves (Ford, Apple, DoD, HP, etc) but for every /8 that is returned by these organizations, it would amount to only a month or two of delaying the exhaustion of the IPv4 space. There’s only at most a dozen of these organizations from what I last recall so it’s not worth it anymore at this stage. We might as well just keep going to IPv6 and benefit from the new features that it will eventually introduce.

  4. Nice post Jose. Didnt know this blog existed until I was pointed here by Scott B on Yammer.

  5. Paula Murphy says:

    Great writing Jose, I would never have known of the hidden talents you have.
    This provided a great overview for us lay people and entertaining to read.

%d bloggers like this: