Slashdot Mirror
What's Your Timeline for IPv6 Migration?
SgtChaireBourne asks: "IPv4 has, over the last 20 years, seen unexpectedly wide adoption. During this time it's proven to be both flexible and robust, but also several problems, though once small, have grown. IPv6 looks to solve some scalability problems, add needed privacy and authentication mechanisms, address quality of service, and provide better routing and addressing capabilities. What kind of timeline does your site/institution/business have for rolling out IPv6 and how?" Those interested in IPv6 migration may also be interested in this article, from a year ago.
"Nobody else is, so why should we?"
That's basically the position we've taken for some reason where I work. Sure, we've been toying with grabbing a block and deploying it on some of our core routers across North America, but...there's no real need per se to do a serious deployment. Nobody's been asking for IPv6 either.
Maybe if there was a way to have mandatory conversion, things would move along a lot quicker.
In today's business climate, we can't imagine migrating without a financial incentive to do so.
IPv6 is like BetaMax tapes back in the 80's: sure, the format is technically better, but we've already got a ton of IPv4 gear and software. Even if you only use free software, there's still man-hours involved for implementation and planning. I pity the fella who walks into his boss's office and says, "Yeah, I'll be spending the next week on the IPv6 migration, getting all the desktops working, upgrading our router firmware, getting an IPv6 address from our ISP, etc."
IPv4 will work just like VHS tapes did: it'll be fine until the next dramatic quantum-leap comes along, like Tivos and DVD recorders will cut down on VHS recorder sales. IPv6 has some neat features, but nothing that a typical small business can't live without.
In the go-go-90's, you'd have been able to pull it off, but these days, if it ain't broke...
What's your damage, Heather?
I've already switched, but isn't it more important whether all the really huge backbone servers switch? I mean, the majority of them are using IPv4, so are they willing to shut down for a few moments to upgrade (assuming it takes that long)? If they switch, could that entail major loses in their companies income?
IMHO nothing too bad with IPv4 and NAT... if it was implemted properly.
Instead whole blocks are hoarded and even using NAT becomes hard.
What about dynamic IP? So IPv4 or IPv6 as the base, but a free adotion of freely routable/accessable levels below this? I can imagine if I get the 'dream' of a directly accessably washing machine, fridge, curtains, etc etc etc I'll need a whole lot more exernally accessable addresses.
So I think: either a standard port routin for each appliance under IPv6 or a dynamic range under the UPv6 range.
I think IPv6 only delays the problem.
IPv6 Should be built alongside and parallel to current Inet. If it is done parallel to the Inet, we could fix alot of what is broken with the Inet.
Addressing is just one of the issues that IPv6 addresses, but the Parallel nature that I am proposing would fix things like Security, Spam, Porn, Enum, Virus, Streaming media, meta port assignments, directory services etc.
There is much more. Trying to build IPv6 ONTOP of the current Inet is just as broken as the current Inet.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Such a huge update would mean the end of anything less then WinXP in the Windows world, you aren't likely to see many companies completely upgrade every machine in an organization to WinXP until there is a business need, other then just being ready.
The widespread use of NAT and RFC 1918 address space has somewhat mitigated the need for more address space. I realize there is more to ipv6 than just more addresses, but I think shrinking ipv4 space is going to be the thing that makes everyone switch over.
I'm thinking that it'll really start to get to the point where I will start using it in 2008. This is me speaking about my small hosting business as well as a member of the local Internet Cooperative. I'm sure I'll be playing with IPV6 in the next year or two to get up to speed on it.
At the moment you can't get IPV6 service from any of the large providers. And really only people on ipv6 can take advantage of it, so... Until a significant portion of the end-users have IPV6, I can't see that we'll have any real need to start using it in any real way...
It's, obviously, a chicken-and-egg thing. It was really pushed because of the "sky is falling" shouts about running out of IP space. Todays world seems like there's plenty of IP space, if you're not super wastful with it, and we have other problems to face like router table space and ASNs.
The other problem I don't think we really have ironed out right now is that the routers are really underpowered and optimized for ipv4 routing. I expect that having significant traffic on IPV6 is going to stress many of the bigger routers on the net to the point that they can no longer function. Lots of "big router" admins are already working hard getting the routers to handle current traffic.
Sean
Well put. Thats what I see from all the companies I consult with. Don't hold your breath. The cost/benefit just isn't there, and won't be for the forseeable fututre, i.e., years.
No kidding. Switching to IPv6 is more work than it is worth right now, and the balance doesn't seem to be shifting. I remember about 5 years ago when a Prof told our class that we should invest in Cisco because everyone was going to have to replace all their network hardware to switch to IPv6. I've been following the stock since then for kicks, and you would have been much better off burying your money in the back yard.
Besides, we almost HAVE to use NAT to prevent p2p apps from completely swamping our tiny college connection, so we have unlimited IP addresses anyway.
Multicasting isn't an IPv6 specific feature. It is available in IPv4 as well but there isn't wide adoption of it throughout the Internet yet. Switching to IPv6 is not going to change that by itself, you'll need to persuade the ISPs that it's financially worthwhile to route multicast traffic...
I've seen no plans to migrate to V6 here either. The routers can go with just a firmware upgrade I understand though, but the bottome line is WHY ?
With NAT I just don't see the need. At home I'd rather not have a bunch of registered IP's for everything at my house anyways, make the stuff work behind my firewall, anonymously
errr....umm...*whooosh* *whoosh* Is this thing on ?
Once IPv6 has a killer app, you will see widespread adoption. Until then, who really cares? There just isn't a real need for it.
Nobody -- not ISPs, not users -- is going to switch to IPv6 until they have a reason to do so. Private networks have obliterated (not just mitigated, in my opinion) the argument that IPv4 does not offer enough IP addresses for everyone. We have all the IP addresses we will ever need using IPv4 and NAT. That was once considered the main reason for IPv6 adoption. Now there isn't much of any reason to switch, other than the coolness factor that only techies will appreciate.
I'll roll out IPv6 as soon as there's some pr0n on it that I can't get via IPv4.
.sig
-- this is not a
So far over the horizon that its dropped of the radar screen. I think most organizations have this on the back-burner if it has been thought of at all.
Ironically, the best feature of IPv6 is also probably its Achilles heal. Until corporate isp's figure out how to price Multicasting without shooting themselves in the foot, IPv6 will never make it to the forefront.
At that point, Marketing is going to turn to Management and ask "Why arn't we using this next generation networking technology?" To which Management is going to go to IS, and ask the same question.
IS is going to report the following.
Management is then going to ask "How long it will take to deploy?", and "How long do you expect to continue working here?". At different companies different emphasis is going to be placed on those two questions.
ISP's and CoLos will have the same set of problems. Large businesses are going to ask why they are not ready for IPv6, and will have to seriously look into how much longer it will take before they start loosing their big customers.
At that point, IPv6 will be discovered as already existing in just about every router and server OS that is out there. The exceptions will be hardware that is due for replacemnt shortly anyway.
People who have been fighting with silly problems with IPv4, will crack open the manuals on IPv6 and realize that almost 90% of the problems they have been fighting with, dhcp, ddns, IPsec, IPNat, are already built into the technology that they already have deployed and mearly need to add a few statements to interfaces on routers in their network.
The early adopters are going to move their CoLos out of the US to countries where the CoLos have already deployed IPv6 in their infrastructure. Some of them will prosper on the added business, some will not get it right and will fail.
Nay-sayers on Slashdot will point at the failures in the early adopters and say "I told you so, the technology ain't ready."
Are there problems with the above senario? Sure. There are problems with some of the deployed IPv6 stacks on some Cisco routers. There are questions about the efficacy of using some of the applications that businesses are using on IPv4 being migrated to IPv6. I understand that there are Novel 3.2 servers out there that are still in use because the company using the server has a functioning solution even if spport costs in the future are going to skyrocket.
Those of you complaining about being out of work, might want to spend some time at the library and brush up on both your IPv4 and IPv6 knowledge. You will then have a potential advantage over those people currently working, fighting with IPv4 problems and ignoring the possibility of using IPv6, because "No one has found a real need for it."
After all, I could be wrong.
-Rusty
You never know...
IPv6 is like BetaMax tapes back in the 80's
As with most attempts to use the BetaMax analogy in the computer world, this one fails: BetaMax was incompatible with VHS, period, end statement. If you had a Beta machine, VHS tapes were useless to you, and vice versa. IPv4 and IPv6 can happily co-exist, though. Totally different situation.
That said, I agree with the underlying premise that migration isn't going to happen until it's easy and cheap, and (moreover) there's some motivation out there. It's possible that this translates to "never"; it's also possible that it translates to "some time in the next 5-10 years". I'm reserving judgement for now, but I'll be amazed if I have to deal with IPv6 in less than five years.
That's when i'll change...
and I dont think it's that far off - AOL probably have more need than most and might pull it off more easily.
They already need heaps of IP addresses for all their dialup users.
Most aol users wouldn't give a monkey if they installed AOL v19 and suddenly it used ipv6... they just wouldn't notice.
The remaining computer literate aol users (if they exist) would probably be quite pleased.
Just my thoughts.
Uhm, hello?
:-)
IPv4, a.k.a. "the Internet" has seen an unexpected adoption in terms of world domination. You know, the reason that you're able to make this comment. If you thought that part of the story already was about IPv6, well, read it
"We can confirm that Debian does *not* ship the version with the trojan horse. Our version predates it." [CA-2002-28]
The problem I see with multicasting is, you'll probably still have to pay for the bandwidth, because at some point the data still has to be replicated. Do you think that your ISP is going to sit and watch as you multicast through their network, causing them to send out many times more data than is coming in? Not if they can help it. They will charge for every multicasted bit. Maybe it's more efficient sometimes, but it will still cost lots of money, most likely. It's not going to help you host linux isos on your DSL line for your regular flat rate service. At least, not if your ISP can help it.
main(c,r){for(r=32;r;) printf(++c>31?c=!r--,"\n":c<r?" ":~c&r?" `":" #");}
DNS is not always available, and it doesn't really answer my question, does it?
If tits were wings it'd be flying around.
It isn't really the service providers, but software providers. The tier-1 ISP's will not be able to run native IPv6 until their software providers for vital management tools (Concord, Micromuse,HP,InfoVista,Lucent) can provide the support. It would be impossible to manage a network in this time without quality management and reporting tools.
I cannot imagine that UUNET or a similar provider will move to IPv6 before they have the ability to manage it at the same level as they do now. Certainly the Tier-1's can make the decision to go, but not until their software can handle it.
Had IETF chosen to set aside of chunk of address space to permanently and portably allocate to serious deployers ... space that would not ever be taken back ... that could be kept forever as the payment for helping to make IPv6 happen ... then I think a lot of ISPs and businesses would have done this. Instead, what we have are 6bone addresses that will not be routable on the real IPv6, and tunnels that will be taken down soon, making those addresses useless. Sure, there is a routing scalability problem still in IPv6. The only benefit IPv6 has over IPv4 in routing is that there hopefully won't be a case of single companies advertising dozens of unaggregated prefixes ... or at least no more than one per major location. So shame on the IETF for not having solved that problem with a fundamentally new way to do routing in conjunction with the development of an addressing technology that now way overscales the ability to route it.
It's now a chicken and egg problem. ISPs simply will not, not in this economy, and not for years even after it gets better, make an investment in deploying IPv6 unless there is customer demand for it. Customers won't demand it until there is some real need for it, which is not the case, especially with so many businesses now running big LANs via one NAT'd IPv4 address. If some web site goes online with both IPv4 and IPv6, everyone will access it via IPv4 and that won't create any demand for IPv6. If they go online with IPv6 only, no one can reach them for a while, and they will probably not really make it.
But there are some possible ways to make IPv6 happen:
now we need to go OSS in diesel cars
This is slashdot. Microsoft gets blamed for everything here.
Do you think that your ISP is going to sit and watch as you multicast through their network, causing them to send out many times more data than is coming in? Not if they can help it. They will charge for every multicasted bit. Maybe it's more efficient sometimes, but it will still cost lots of money, most likely.
Naw. They get their money from the people the multicast bit is going TO. Replicating it means more people upgrade and pay for bigger inbound hoses.
Think about it: Got broadband? Didn't you pay a premium to get a fat INBOUND pipe? Isn't your OUTBOUND pipe pinched down a bunch? Don't you use it that way? Do you feel cheated because your outbound pipe is narrower than your inbound? Or do you watch your streaming programms and suck down big images on the web with a few characters of URL going the other way?
Now if you could originate video streams and feed a LARGE audience on a DSL that was good for one stream upbound and several down, and a bunch of others could, too, and these indies made enough programming to convince a few hundred thousand users to upgrade to such fat pipes and pay a higher fee, and the ISP only had ONE COPY of this content-feeding-thousands on any given internal pipe rather than several, do you think the ISPs would nix it? Or would they sell it to all comers and laugh all the way to the bank.
It's CONTENT that drives internet expansion. And right now the general user as content provider can't feed enough people to make it worthwhile. So the main thing that's popular in peer-to-peer content provision is so-called piracy - where a CROWD of people each serve a FEW consumers with content mostly cloned off other people's well-advertised productions.
With broadcast origination available to general users, ORIGINAL content can reach enough people to justify the production costs. Without it, you need a major-league expensive infrastructure even for webcasting.
So the ISPs have a fine financial incentive to allow it once their infrastructure is up to it - and make the bucks back from the increased feed for pipes fat enough to originate and view it.
Or at least the ones that are NOT owned by a media conglomerate do. The ones owned by a media conglomerate have an incentive to suppress any broadcast technology where they don't originate the content themselves - because it represents competition for their more lucrative content-production-and-distribution business.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
The current ATT network was created out of the old ibm as well as other networks, i'm not going to read the entire ipv6 routing table (well, it is short enough to read actually, but i'm being lazy) to check for one of the many ATT legacy ASNs or SBC ASNs that they may be using to operate their IPv6 network. I suggest checking 6bone pTLA listing or with the Regional Internet Registry for people that have been assigned IPv6 address space. In the US at least, it's an InterNIC-type company (remember inernic?) called ARIN
What's wrong with telnet?
Ok, rephrase, other than the plain text transmission, what's wrong with telnet?
Along the same line of thinking, if you want to get rid of telnet, do you want to get rid of FTP? That's essentially what FTP is - unencrypted uname/pass auth. So, what's the difference?
We still support telnet at Netmar, because our users have telnet on their computers. Joe Blow, with his windows 98, can understand "start - run - telnet login.netmar.com". He may not understand "download putty, change to SSH, port 22, type in login.netmar.com, press connect", or be willing to deal with it.
We do, of course, provide documentation on how to do both:
http://guide.netmar.com/connect/command-line.html
But, seriously, telnet doesn't hurt anything. We don't allow it for root, our dedicated server customers must enable telnet for themselves (most of the time we don't even install it, or install it and leave it disabled).
It's the same kind of compromise as the Lindows article today: Security vs. speed. Telnet is fast, ssh is slow. Telnet is insecure, SSH is safer. FTP is fast, sftp is painfully slow.
Stop and think for a minute - All of this unix/linux/*nix/bsd/etc etc stuff that we all know and love is *old*. I mean, the basic premise is ancient, in computer terms. But, it has evolved with the changing times, and quite nicely. Just let it do it's thing, and phase things out when they are no longer useful. Telnet still has a use. IPv4 still has a use. Sendmail still has a use, regardless of what Bernstein may say.
I say, I don't want to give up 30 years of tradition every time something new comes out. Just go with the flow. Let it happen in it's own time.
~Wx
sig?
DJB wrote:
Local IPv6 addresses don't offer any advantages over 10.* IPv4 addresses.
they do, though. having true end-to-end communication means that peer to peer applications like voice over IP or BitTorrent actually work.
It also means that users on larger networks are actually accountable. if you have a way of uniquely identifying a machine from outside the network, abuse complaints actually mean something. if the secret service comes knocking on a network admin's door complaining about threats being sent from your network to president@whitehouse.gov you can't say "oh, we don't have any way of knowing which user sent that mail, because it didn't go through our mail server and all 5,000 machines on this network connect to the internet through the same IP address." chances are that you aren't logging every connection that goes through your nat gateway, and so your basically stuck holding the ball on that one.
Global IPv6 addresses don't work. Most client computers around the Internet can't talk to a server on a global IPv6 address, and most server computers around the Internet can't talk to a client on a global IPv6 address.
of course they do. every host in my home network has a globally routable ipv6 address (thanks to hurricane electric's tunnelbroker.net) and i can reach hosts at my colo provider that are set up via freenet6. i can also reach hosts at my school that are directly connected to the ipv6 backbone via nysernet.
All the operating systems I use have been claiming ``IPv6 support'' for years. But they still require manual action by the system administrator before they can talk to IPv6 addresses.
no they don't. radvd is like dhcpd on steroids. if your hosts are ipv6 capable, start up radvd on your ipv6 connected router and within seconds every one of them will have their own globally unique, routable ipv6 address.
(All of this boils down to a small protocol design error in IPv6. A small change to IPv6 software would make IPv6 addresses work without any administrator action. I have a web page, http://cr.yp.to/djbdns/ipv6mess.html [cr.yp.to], explaining this in much more detail.)
this page basically says two (false) things.
1) you can't use ipv6 and ipv4 at the same time, so if you switch to ipv6 now you can't reach 99.9% of the internet.
this is blatently false and you know it. ipv6 and ipv4 can co-exist on the same machine very well. on my ipv6 enabled network, every host has an ipv4 address from 10.0.0.0/8 and an ipv6 address from 2001:470:1f00:321::/64. my machines try to look up AAAA records on hosts first, and if one exists they try to connect to that ipv6 IP. if no AAAA record exists, or the host is unreachable via ipv6, the machine falls back to ipv4, looks up a host, and connects.
2) it takes a massive amount of work to convert all applications over to ipv6 and no one has even started on such a task.
this one is even more confusing. i've got ipv6 enabled apache, ipv6 enabled qmail, ipv6 enabled djbdns, ipv6 enabled mozilla/phoenix, ipv6 enabled xchat, ipv6 enabled internet explorer, etc. all of these applications on every modern OS have all been written to use ipv6 first, then fall back on ipv4.
Up until last year I have been developping firmware (embedded server load balancing) for the new line of switch/routers of a big company that shall remain nameless.
At no point were we instructed to even think about IPv6. IPs are internally encoded on 4 bytes, in various and sometimes rather obscure locations within the code. They can even be found as fixed-size strings for ASCII representation for command-line processing.
From what I have seen the rest of the firmware, all the way down to the proprietary ASICS's internal registers, it has no provision whatsoever for IPv6.
Given the amount of work needed I doubt this line of switch/routers (edge & core) will ever support it, despite the fact that this is a new product and the next generation won't come out until 2-3 years - if there's any.
NAT is bad. It screws up the end-to-end transparency of the Internet. People shouldn't rely on it as it only delays the inevitable: IPv4 adress space exhaustion. With IPv6 you (as the end user) get a ~2^64 usable address space, aka /64 prefix.
Well, too bad IPv6 is not widely supported. I'd like my ISP to deliver *native* IPv6 services. As of now I'm running IPv6 through a tunnel with all the associated problems: long delays between hops, shitty DNS resolution (for reverse records), etc.
Hm, rahter than using NAT one could use some kind of 6to4 translation and have his network run on IPv6, provided that it's supported by the nodes. Windows doesn't even come with IPv6 out of the box, and Linux distributions are somewhat lacking in the field. The BSDs are way ahead of everyone else I guess. Mostly because of the japs, who afaik are running out of IPv4 addresses.