Slashdot Mirror
IPv6 Transition to Cost US $75 Billion?
darthcamaro writes "There are alot of reasons why the US isn't moving as quickly as Japan and Europe in migrating to IPv6. One of those reasons is likely cost. An article on Internetnews.com cites an unreleased 'Dept. of Commerce report estimating it will take $25-$75 billion to pay for the transition.'"
$50B difference is huge, this goes to show nobody knows what's going on.
I guess USA's high internet adoption and usage actually hinder its move.
This reminds me of China's ability to build its new Shanghai rail based on the magnetic levitation system, while other well-established rail-using nations like Singapore may find it difficult to switch. Talk about right place right time.
Virtual Betting on Facebook for non-geeks.
They should have focused on how it will *GROW* the economy by creating $75 Billion in new jobs and infrastructure.
How much money would be spent on upgrading routers and internet infrastructure anyway? I can claim that over the next 10 years internet infrastructure will cost over $100B, regardless of whether or not it's IPv6 compatable or not.
Since this changeover is going to require something new, does anyone have a list or know of a place that talks about exactly what needs to be done to switch over to IPv6? Like routing tables, software upgrades/changes, hardware upgrades, network changeovers and what else?
Fly me to the moon Let me sing among those stars Let me see what spring is like On jupiter and mars
Is it really worth it?
The AACS key is NOT 0xF606EEFD628B1CA427BEA93A9CA9773F
If we eliminated most of the fraud, waste and abuse in the government. With the Department of Education not being able to account for a majority of its budget, the Defense Department losing over $12B of tax dollars in Iraq and all of the pork that goes through Congress, I can't help but think that if the Congress didn't have the power to spend money on "internal improvements," we'd not be in this problem today.
The governments in this country waste so damn much of our GDP on pure bullshit that if we actually had fiscal responsibility, this would be non-issue. We have a GDP of $11T, does anyone actually think that if the costs associated with compliance, regulation, tax payments, etc. were much easier that corporate America would be bitching about this transition? It'd be just a drop in the bucket.
Click here or a puppy gets stomped!
The Department of Commerce does not have a very good track record of forecasting market trends. I think this report is especially callous as investors might heed this "warning" and invest in IPv6 companies prematurely.
_No one_ knows IPv6's cost. The market will see a few early adopters, then a steadily growing medium-sized business buy-in, followed by a boom of users or a bust due to newer technologies.
For a government agency to print these assumptions makes me think they either needed some media spotlight or the researchers wanted their stocks to go up.
how much of that is money that would have been spent anyway? Upgrading routers and stuff when it's time came around to be replaced - it may take some time, but if all the new equipment bought is IPv6 compatable then it will only actually take the flick of a switch to make the transition - it just may take more time until the complete transition is possible.
Most routers run *nix. IPv6 transition thus requires a "firmware" update.
Not that most equipment companies will do this mind you. They are quite happy having you purchase brand new networking equipment.
not to make the transition?
That's a lot of bucks, but studies like these are easy to take in isolation instead of looking at the big picture.
The U.S. economy is what? About 12 Trillion dollars a year? In 1999 the internet economy was closing in on 150 Billion, by now it has to be through the roof.
Poor software? It costs over 200 Billion a year (sorry no link). You have to put these numbers in perspective. When you are dealing with 300 million folks or so, and the world's largest free market, any kind of estimate for anything is going to be big. The common cold costs over $30 Billion a year.
Just keep it all in perspective. The internet economy will blow right through this obstacle if it gets in the way of sales
My Blog
Essentially it means increasing the length of IP addresses (because we only have a finite number and with IPv6 they claim there's enough IP address for every grain of sand on the planet) and IPv6 includes more information in the header to allow for better efficiency / functionality. As a result of this you'd need new routers, switches, and possibly network cards. The new routers and such would have to be IPv6 and IPv4 compatible because you can't switch them all at the same time. Obviously, it will cost money to make the change, but I think the $25-$75 billion is just plain made up to try to scare people.
IPv6 isn't going to happen, because it doesn't need to happen. I can get to all the web sites I need. So can you. My coffee pot isn't on the internet, and if it needed to be, I'd use NAT, or invent some new multiplexor protocol that sits on IPv4. Don't you people realize this???
I love the guy up their who said IPv6 will *create* $75billion in the economy. How the hell will it do that?
I'll issue my usual challenge to the IPv6-fans: If you love IPv6 so much, cut yourself off from IPv4 completely. Don't use an IPv4 address. Don't access IPv4 sites. That's what has to happen for IPv6 to be a reality. If you're running IPv6 on top of or alongside IPv4, you haven't "switched over" yet. You're just goofing around with some nonstandard network protocol. Might as well use fidonet.
Go ahead, I'm waiting......
...and therefore assumes that it will be carried out under a no-bid contract awarded to Halliburton, who will bill Cat-5 cables at $10,000 each. Sounds a fair estimate to me :-)
Also, have you heard of: "::aaa.bbb.ccc.ddd"?
What previous version of IP are you talking about? You aren't seriously referring to Arpanet's NCP to IPv4 transition in 1981-1982 are you? Arpanet had roughly 200 hosts back then!
Uh, what previous versions of IP? The Wikipedia article on the Internet Protocol says "Versions 0 through 3 were either reserved or unused". And even if there were versions before IPv4, the deployment would have been what? a dozen machines? We're talking multi-user mainframes and mini-computers at universities, not home PC's. Nothing like the hundreds of millions of hosts using IPv4. Why would the designers of IPv1 through IPv4 have made the addresses a superset of the previous version, when so few hosts actually exsisted? Why go to that trouble when it wasn't really a problem?
We must move to IPv6, because the Internet just doesn't seem to be working right (or at least I tell myself that, because I wouldn't want to fix it if it weren't broken). I look forward to a time that each of my Happy Meal toys will be able to be connected to the Internet, yes we need IPv6 Now!
Bah! As others have pointed out, there will not be much cost, if it rolls out more slowly. As you update hardware, get stuff that can do both IPv4 and IPv6 next time... eventually a critical mass will be reached and the switchover will happen.
This issue is a bit more complicated than you think.
And we don't have to wait for our ISPs, either. I've been using 6to4 (IPv6 tunneled over IPv4) for years. It's especially useful on home networks where multiple servers have to share a single IPv4 address on a cable or DSL modem.
6to4 works very well. A 6to4 tunnel coexists nicely with an IPv4 NAT on my home router. The computers on my home network can run conventional clients through the NAT just as they always have, while servers running on those computers can be contacted directly from the outside using IPv6.
While not every Internet application yet speaks IPv6, the important ones already do. SSH is the most important, but popular SMTP, IMAP and HTTP implementations do as well.
I cannot believe the handsprings users are expected to perform on retail commodity routers with kludges like "port forwarding" when 6to4 tunneling is both simpler and far more general and powerful.
I'd guess that it'd cost you a couple billion dollars getting the really big peers using IPv6 between each other and just tunneling IPv4 traffic around. By the time you get the players to the table and you get Cisco, Juniper and the big consulting firms all into it and the couple really big ISPs to show up at the table you've already blown a few million. By the time the pissing contests are all done, I could see that "project" costing a couple billion, easy. It's going to be a money grab any way you cut it.
When you step back and look, you can migrate it down, start at some top level peers and push it down. ISPs can roll it out, smaller ones would have it easier but it's doable. A couple of big schools could switch over to IPv6, there is enough support between Linux, Windows and MacOSX that all the students could show up one year and it could be an ipv6 network. You start looking at businesses and that's when it starts to look like a mess. Every layer 3 device in the business will have to be evaluated, many replaced, many might not even be really audited as it is. It looks like y2k all over again. I don't even think most IT guys have IPv6 on their mind when they buy stuff.
I'd wager, you could get all the top level peers to IPv6 and that would cost $5B or there about. Maybe you could get the 10 biggest ISPs to switch, that'd probably cost another $20B, I'm just swagging $2B a pop for the big 10, probably lot's of infrastructure upgrades to make it happen. So that's $25B. Start looking at corporate networks, who knows? I'd imagine it would cost a pretty penny just to get them peering IPv6 and using IPv4 internally; of course by this point there might be some really good reasons to use IPv6 across the board because you'll be losing access to things from IPv4.
You take another step back, I setup IPv6 at home for my own good, not super easy. All my network jock buddies are IPv4 savvy and while they know specs and such, none seem to have a lot of IPv6 known-how for real. I think there is a pretty big void there. On the up tick, I think this is going to happen, it's just a matter of when, I also think that if we play our cards right we might be able to clean up a lot of the bullshit that is out there right now.
Huh? RFC 4038 says this:
This seems to imply that IPv6 does contain the address space of IPv4.
Of course, for it to be useful, eventually people will have to start using other addresses that are not part of IPv4. But no protocol design can get around that problem.
You also need keys to all the offices where there are desktop machines that have static IP addresses. Or any desktop machine that can't be automatically remotely reconfigured to use IPv6. Hey, speaking of that, is someone in some dark corner still using Windows 98 or Mac OS 9 for something important? Probably. Do those support IPv6?
Oh, and by the way, what about the downtime of the servers as they are restarted? And do you think you'll be able to complete that firmware upgrade on the Cisco router and reconfigure the router with no downtime? If not, how are you going to ensure that the entire network being down doesn't cost the organization money?
I've tried to get this point across before, and I'll continue to. The transition from IPv4 to IPv6 will indeed cost quite a bit even just for upgrading hardware. Here's why:
The transition will happen incrementally and will involve a period of IPv4 and IPv6 overlap where any given internet entity needs to have both an IPv4 address and an IPv6 address--obviously, most will have plenty more than a single address. Presumably your browser will then attempt a DNS query for the IPv6 address and fail over to IPv4, though some may give priority to IPv4 for various reasons. The overlap may be in terms of hours or it may be in terms of years for any given entity. We have the means to translate the entire IPv4 space into an IPv6 range, so many people won't notice when their ISP switches.
The cost, aside from technical hands to make the transition, will come in terms of hardware. If you run your router with 20% free memory and 70% CPU load, then you are in the green by today's standards, but when you add in IPv6, you need to upgrade memory and you start dropping packets to load. This doesn't take into account any extra fun that your router may need to do.
I don't know what percentage of active switches do IPv6, but many of the older switches will start broadcasting the IPv6 because they don't understand it. Now your switch is a hub. That will choke any network that needed switches in the first place. Even a single switch that doesn't do IPv6 could take out a large chunk of your network.
So to sum it up, the transition period will crank up the router requirements by 150 to 200% of the current requirements, which will blow much of the internet into the packet-dropping red. Many switches will need upgrading, which may include hardware. Network techs will be paid for all of this. Don't forget that any company that plans to have continued coverage will also need to have an escape plan in case things don't work like they are supposed to.
First off, I'm not sure that there's more households on the net in Europe and Japan than the US. I'd want to see a reliable source on that. However all that aside, that's not the real problem, companies and major backbone infastructure is and in that the US is far ahead. Though the Internet has been growing globally at a fantastic rate, espically since 2000, it it is still heavily US based. There's a lot more large (and thus expensive) infastructure that needs to be upgraded.
Also the US has an additonal problem in that a lot of the net here was developed before other countries started really taking off. That translates to old equipment, some of it has been replaced, of course, but an amazing amount of it still works fine and thus is still in place.
Well, when you are building new infastructure, it's not that hard a sell to pay whatever small amount extra to get IPv6. When you already spent $500million on a facility that still works fine, hard to get another $500million to upgrade it for a feature nobody is really asking for yet.
A few simple classes of automatically granted prefixes would do wonders along with a few simple routing rules. For example, to accomodate situations where NAT is used as a workaround to scarcity rather that for security, why not an address in the form of prefix, IPv4 address (of NAT device/firewall), arbitrary Ipv4 address. Addresses in that prefix could automatically be encapsulated into IPv4 to the firewall. The source address would make the right thing happen for the reply.
/80 range to play with. No provider support needed.
Doing things like that could create an incentive for the edges to upgrade even if their providers can't be bothered to do so. It would provide value even if individual workstations/PCs were upgraded (OS) for IPv6 support. Intermediate routers wouldn't even know the difference.
What you describe has been available for *ages*. It's called "IPv4 mapped IPv6 adresses", and it basicly boils down to the entire ipv4 range being a subrange of IPv6 (From the top of my head, the 3ffe:: is used for this). Still, it hasn't quite caught on.
FYI: The opposite is available too, in a way. It's called 6-to-4, and enables *everybody* with an IPv4 address to use IPv6. You even get an entire
It would provide value even if individual workstations/PCs were upgraded (OS) for IPv6 support.
The majority of the workstations/PCs doesn't even have to be upgraded. Windows XP natively supports IPv6 (it's merely disabled by default, but a simple 'ipv6 install' on a command line fixes that), Linux has been having support of it for the last few years, so unless you're running an ancient machine that would need updating for several other reasons, consider yourself IPv6 capable. MacOS X has been having IPv6 support since the early OSX days, IIRC. The list goes on and on.
Heck, even Internet Explorer and Firefox can use IPv6. What more client support do you need ?
IPv6 users will only be able to use "passive" protocols (say goodbye to DCC transfers on IRC, and lots of multiplayer games)
Bullshit.
on the contrary, IPv6 will allow multiple people behind that firewall to have dcc transfers, and do ftp much more easily. no more ftp or irc modules required to nat shit around, no more H323 gateways and gatekeepers needed because person A in his 192.168.0.foo network can't connect directly to your 192.168.0.foo machine on your own network (duh, to his H323 client, this would be like connecting to localhost)
Lastly, who wants to remember an IP address like 2001:0db8:85a3:08d3:1319:8a2e:0370:7334/64
guess you ultimately don't know jack about what you think you're talking about.
hey, dude, that's what DNS has been designed for.
see, if your computing skills were about half of what your BS-spouting skills are, you would have noticed by now that the docs refer to AAAA records to store that info for you. besides, the router affecting the IPv6 addresses with dhcp would also serve as your dns server, and store that info. Whenever your box asks it's IPv6 address to the router in question, it says
machine1: hello, I'm foo.bar.baz, what's my IP ?
router: ok, foo, your ip is 2001:0db8:85a3:08d3:1319:8a2e:0370:7334/64
now, machine2 wants to communicate with machine1
machine2: hello, it seems you are the manager for bar.baz, what's the address of foo ?
router: erh, lemme look... there it is. foo is 2001:0db8:85a3:08d3:1319:8a2e:0370:7334/64
now, can you explain where you see a human being remembering the address in question ?
on the contrary, IPv6 will allow multiple people behind that firewall to have dcc transfers, and do ftp much more easily. no more ftp or irc modules required to nat shit around...
The whole point of a firewall is to drop packets that could be malicious. Just because all the computers on the network have their own public addresses doesn't mean you should pass along everything that gets sent to it.
[regarding long IPv6 addresses] now, can you explain where you see a human being remembering the address in question ?
If you've ever tried to troubleshoot DNS problems you would know how important it is to know important IPs. AAAA records just aren't helpful when the DNS server is broken.