All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Combatting IPv4 Depletion and the Merits/Drawbacks of IPv6 as a Solution
A continuation of this thread that went way off topic. See page 3 in particular.
http://lowendtalk.com/discussion/40680/chicagovps-spammers
@bsdguy said:
Hoster: You can have 2 million IPv6 for free!
Customer: How much is an extra IP4?
Hoster: 2$ per month. But you can get yet another 2 million IPv6 for free for a total of 4 million IPv6!
Customer: OK, please add 1 extra IP4. I'll pay the 2$/month.THAT's a typical conversation today.
Funny thing: If IPv6 is the best thing since slice bread, why then did most telcos und ISPs take up IPv6 only hesitatingly? And many still don't talk about, even less recommend it.
That's important because the telcos/ISPs are basically the only way to rape and force feed IPv6 down customers throats. And they seem not too eager to do that. Of course, they'll also be the idiots who will have to pick up the costs (billions of $ globally) for support. Because, guess what gazillions of Johns and Maries will do? They'll ring the support lines.
I never said that IPv6 is the best thing since sliced bread. It is far from that, but it is the best solution to the IPv4 depletion problem.
Telcos and ISPs are stepping up to IPv6 because they realize that they have little choice. Do they like it? No. Will it cost a lot of money? Yes. Will they have to support their customers as they migrate to IPv6? Probably, and it won't be pretty.
Is there an opportunity for those who want to make IPv6 plug and play for John and Marie? You bet. Will it get easier for John and Marie? Yes, over time.
Could we have done better? Yes, but it is too late for that.
I still maintain that IPv4 will become more and more problematic, and everyone will coalesce around IPv6 as a solution.
Comments
Front up: I didn't mean you in particular.;)
As for the matter: "best solution to the IPv4 depletion problem" - I doubt that. In fact, I even doubt that there is a IP4 depletion problem or, more correctly, a real depletion problem.
Why? Simple. Assuming that each and every human (about 7.5 bln) has his own DSL plus a smartphone, there would still be no problem if - and that's the true problem - a very considerable part of IPs were not - completely needlessly and thoughlessly and ignorantly - taken away by Pepsi, .mil., .gov, .edu, etc.
Using NAT, a solution that is well established and globally available, we could easily connect all those home boxes, company boxes, etc. and all those smartphones 50 times over - and still have 2 bln public IPs for servers.
Unlike some here that's what I'd call a good solution. One that satisfies all demands easily and doesn't need any major changes.
As for large corps, agencies, they have to manage their gazillion public addresses today anyway. And if they are halfway reasonable and professional they assign internal IPs (public or private, no matter) through DHCP anyway. Looking at it pragmatically, it'd cost them some weeks of planning an then the major part of the change could even been done scripted.
Note that I do not want to for IP4 eternally. Sooner or later we will want something bigger, no question. But then a sensible and well thought out 64 bit based solution will easily do. After all that'd be about 2 billion IPs for each living human.
From an engineering and design standpoint, it's a really shitty solution. But it's the only one we have, nobody bothered to come up with anything better. We're stuck with what our laziness hath sown.
I work in telco. It will be a very long time until IPv6 is deployed on a scale that makes any difference. Sadly . . . sort of.
Definitely, but we're probably a decade way from ubiquitous IPv6 deployment. There is still lots of IPv4 space being held in reserve by assignees. IPv6 will only become ubiquitous when the opportunity cost of not having it exceeds the financial cost of implementing it.
One way to tell how IPv6 is doing is to go to bgp.he.net and compare:
Since a /32 turns into /48s turns into /64s per customer; IPv6 announcements should be much more than that of IPv4 announcements, had they been deployed equally. Of course, you can also compare the routing figures; or the total networks figures to come to the same conclusion.
IPv6 simply isn't available enough places I want to go to on the internet, if I were IPv6-only, I could not even use these sites.
@bsdguy: Permit me to sum up your points, and then please correct them where I am wrong:
@emg (me) says:
Okay - let me clarify:
First, I agree with many of your points about drawbacks of IPv6. It could have been done much better. That's true of most things designed by committee. Nonetheless, it is our only real, tested, working option. (And yes, it does not work well. And yes, it is not widely enough deployed for us to know what other pitfalls are in store for us.)
I believe that your proposal to recover IPv4 addresses and NAT everything is a poor choice; a backwards step. It won't help for long enough to do real good, and the pain and cost of doing it would be far worse than getting everyone to IPv6.
Despite the drawbacks of IPv6, it is clearly the way the world is moving. According to Wikipedia, 5% of Internet traffic to Google is now IPv6, and I predict a "hockey stick" curve for IPv6 utilization in the coming years.
We already have hardware, stacks, etc. built for IPv6.
IPv6 was hard enough to achieve consensus at IETF, IPv(next) would be worse.
I am not "pro-IPv6". I am anti-"anything else that is worse".
I dunno. My home AT&T DSL claimed to support IPv6 two years ago, although I could not get it to route past the gateway. When I switched to TimeWarner Cable, IPv6 just worked. It was as close to plug-and-play as it gets. (I did not switch to TWC for IPv6, by the way.)
I think this is where (@Microlinux and @bsdguy) vs. @emg (me) disagree. I believe that at best, we can get only short term relief by recovering reserve IPv4 addresses. I also believe that the opportunity cost of not have IPv6 will exceed the financial cost of deploying it faster than people expect.
We will find out soon enough.
Since a /32 turns into /48s turns into /64s per customer; IPv6 announcements should be much more than that of IPv4 announcements, had they been deployed equally.
Noooooo... I wonder if you're being really that delusional here, or trying to skew things on purpose. With IPv4 providers announce lots of small-ish prefixes they were able to scrap together and/or buy from various sources. That ends up in lots and lots of IPv4 subnets announced. But with IPv6, they just announce the /32 or even /29 they were assigned, and that's it. For most providers that single IPv6 subnet announced will allow them to connect more customers than they will ever possibly have. So the number of IPv6 subnets announced will always be an order of magnitude less than the number of IPv4 subnets. That's also one of the benefits of IPv6, simplifying and reducing the size of the global routing table.
I only have one /64 on my dedi, but 4 users. Sure, a /64 are a lot of IPv6 addresses, but major websites (mail servers) treat a /64 as one user. Looks like I need 4x /64 and i read somewhere that Hetzner only has 65k /64s for all their customers -> not enough.
Just waiting for IPv8 now. -.-
That's not correct.
They have 34359738368 of /64s.
There's something that I almost aggressively dislike, namely the repeated over and over again dogma of the very basis. Which is simply false.
IP4 isn't a bad design. Hell, the whole Internet as we know it is built on it. The, oh so hefty "problem with IP4" can be tracked to basically one issue: Insane, unjust, grossly biased, ignorant distribution.
The rest like IoT, networked devices, fridges, and shoes, mobile IP and the like is simply PR blabla.
In the long run we will need not something different but simply a bigger version of IP4. Maybe we'll travel to other stars, stuff like that. Then we'll need yet more IPs.
But as long as mankind stays below 16 billion people the following is true:
16 bln people ~ 34 bit. Now being very generous and allocating 2 bln IP4 addresses to public servers and another 2 bln. to POP (DSL, UMTS, ...) we can have 16 bln people connected to 2 bln POPs using lousy 3 bits out of 16 port bits and still have each of those 16 bln internet users have several 1000 devices connected or services running, or ...
Short: The oh so urgent problem simply doesn't exist if for 1 and 1 reason only: (mostly usa) corps, agencies, etc. WASTING insane amounts of IP.
Simply forcing them to give up those IPs except for a reasonable but still generous amount of IPs solves the problem for at least 10 years. Simple as that.
During those 10 years - which is plentyful time - we can extend IP4 into "IP5" with 64 bits. Actually, 64 bits is so insanely huge that we could include the 16 ports bits in it, leaving 48 bits for IP which is still a generous 65 thousand times of what we have today.
Sure enough all those IT corporations, that are anyway more marketing blabla agencies than tech companies, will find plenty other ultra-total-urgent gadgets to sell. IPv6 is handy for them but they won't die off without it. After all, there's still gazillions of shoes and handbags without an administration interface to control some blinkenlights.
And replace all hardware, all routers, upgrade all OSes again. Remember, you can't use your "extended IPv4" unless all of those support it. And all for what? A crappier, less advanced version of what we already have in place (every modern networking stack already supports IPv6).
It always cracks me up when some people argue about IPv6 as if we're back in 1996 and IPv6 is still some experimental draft proposal, the "merits and drawbacks" of which have to be considered, and that if something is not to their liking, it can simply be thrown away and we can start from scratch.
Nope, IPv6 is already here, it's already being deployed full steam, nobody is going abandon it and simply write off all costs and effort already spent on implementing it (especially for no show-stopping reason whatsoever) -- so better get on with it or get left behind.
...
inet6num: 2a01:4f8::/29
netname: DE-HETZNER-20071010
descr: Hetzner Online AG
oh cool
Maybe I'll ask them for a few 100 then
You should ask for a single larger routed subnet, perhaps a /60 (16x/64) or a /56 (256x/64). That's the proper way to go about it (and places less load on their routers than a hundred of individual /64s). But unfortunately I also have heard rumors of Hetzner not exactly having their head pulled out of their behind [yet] when it comes to IPv6, so they might not be forthcoming about providing you with any of that.
@rm_
"It's here. Eat it. Period." is not an argument. That very attempt to force-feed and rape people into IPv6 is one major reason why many dislike it.
And no, there would not be a need to again replace everything. As I've explained already a smart approach - unlike IPv6 - would avoid that by mapping IP4 into its space.
Example: If the first x bits are 0 then the lowest 32 bits are an IP4 address. Simple as that. In fact that would be a major argument for "IP5"; it would be a reason for people to really want it and gladly accept it and actually unnerve their ISP to get it.
Thought out and designed properly along professional criteria rather than corporate interests, there would be an easy migration path. One could even with very little effort make POPs/DSLAMS and the like programmable or self-configuring.
Furthermore and importantly, at customer premises could actually keep IP4. The CPE router would simply map it back. Because "IP5" would just be a bigger space IP4.
Certainly, it's deployed here and there. But at the end of the day, we're a long way off from ubiquitous usefulness.
To clarify, I'm talking about allocated space not yet in use - not truly "reserved" space. There is a huge amount of IPv4 hoarding going on.
oh
Which is exactly why no one uses IPv6, right?. Some of my VPS's, my home internet, my 3G, none has IPv6. No IPv6 glue records, no IPv6 rDNS(1), no IPv6 only TOR, no IPv6 Netflix, bing, yahoo, mail servers, non working IPv6 setups (need to ping Hetzner IPv6 gateway after reboot or it won't work), some people only get 1-10 IPv6 addresses out of a /64, which makes all users "share one IP" ...
(1) even Cloudflare did not have working IPv6 rDNS until I asked them to add it because of ISNIC regulations
https://en.wikipedia.org/wiki/IPv6_transition_mechanisms#Stateless_IP.2FICMP_Translation
And yet uptake is poor. And was from day 1. But hey, maybe just another, the umptieth "IPv6 day!!!11!1" will do the trick.
I'll let you in on something private: IPv6 is rotten mainly because of all the corporate interests ("committee" is usually basically just a euphemism for "corporate interests").
And they just love to talk about, invent or plainly lie about "market forces" (allegedly pushing them, for products, for standards, etc). Well - that's the funny part for me - they seem to have missed the market in "market force", namely gazillions of people, many telcos, and many ISPs and other players. Hehe!
At least the uptake for the IPv6 is better than the IP5 idea of yours, which has the same weakness of not being compatible with current hardware and software of today.
ISPs should just sort out IPv6 for customers (my ISP doesn't support IPv6) and then stop IPv4.
When sites realise a massive drop in traffic, they'll make IPv6 available within 24 hours.
It's embarassing how far behind ISPs are with IPv6 adoption.
Luckily, you can run an IPv6 tunnel on any half-decent home router, to take care of this yourself. I'm using HE, which has quite a few free tunnel servers around the world: https://tunnelbroker.net.
What ISP? BT will have ipv6 in late 2015 with a /64 for every customer. Business customers already have /56 and their network supports it fully.
You need BT home hub 3 upward .
You'd be surprised how many ISP's in the U.S don't support IPv6 yet. Cox Cable claim to have "plans".
I configured a HE.net Tunnel in my router and decided to just forget about it.
First of all, I don't think "stop IPv4" will happen quickly. The problem is that people will need dual-stack for a long time, because they won't be able to reach one special, all-important IPv4-only site.
You made a great point that I hadn't thought about. Local ISPs are trying to monetize customer activities on the Internet. For example, note the super cookies that Verizon is using and AT&T tried. If customers resort to various IPv6 tunnels, then ISPs lose the ability to track customer activities.
At the moment, those tunnels are not secure, but I suspect that ISPs lack the deep packet inspection tools to track customers who are tunneling IPv6 traffic. If the traffic becomes opaque (encrypted) then there is even more incentive for ISPs to provide native IPv6.
In other words: Rape and force feed them!
That attitude is massively smart because every ISP, hoster, and web site owner could switch within a day, yeah. As everybody knows, IPv6 is not only heavenly honey but can also be activated by simply clicking "Yes! Hurray! Enable!" in your router. - "Them IPv6 refusers are just stubborn basterds! Probably gay communist muslims or somein, ya know. Hit em and hit em hard. Throw a couple grenades at em, too, just for good measure."
Funny. Switching their internal networks to NAT and private IPs can't be demanded from corps and govs. Too troublesome, too expensive. But asking them to switch to an entirely different protocol and lots of new boxes/costs is fine. Brilliant logic!
I think I'll fly that as my sig for some time
Your're welcome. And so is the honesty of your sig.
@bsdguy has some excellent points, but seems to ignore that IPv6 was developed with more than just IPv4 depletion in mind.
People move on to new standards all the time and it is not always easy. Sometimes government regulations force new standards on people. In the US, we moved away from leaded gasoline and imposed catalytic converters, both for environmental protection. Government regulations specified narrow fuel openings (and spouts) to ensure that new cars could not use the old leaded fuel. We changed television broadcast standards, which imposed a significant change and expense on everyone, but as time progresses, many (most?) of those decisions seem to have been for the common good. In every case, there were protests that the changes were the work of child molesters from hell who were allied with evil corporations with nothing but profit in mind. There is a certain truth to some of the conspiracy theories, but I believe that much of it is for the common good, too.
Standards help all of us. Do you really miss 8-track tapes, which were supplanted by cassettes, which were supplanted by CDs, which were supplanted by digital files? The lack of common standards can hurt, too. Remember the Betamax vs. VHS wars? It took time, and many people believe that the poorer format won (IPv6 vs alternatives?), but standardization on VHS seeded the nascent video rental industry and spurred major growth in the film entertainment industry (who were dragged into it, kicking and screaming!).
To me, the migration from IPv4 to IPv6 is not all that different from the many other changes I have seen in my lifetime. If @bsdguy would set aside his anger and frustration at those big corporations who have wasted so many IPv4 addresses, perhaps he could see the additional values and opportunities that IPv6 brings to the table. It isn't only about IPv4 address depletion (although that was a major driving factor). It is also about routing, bringing back the non-NAT'd Internet, and many other factors.
IPv6 is not the perfect solution. Progress rarely gives us a perfect solution. We undergo evolutionary change.
Here are examples of where we have not been able to change, which frustrate me:
I can understand most of your points. but ...
Great. But the self made "IP4 depletion" is the reason why IPv6 can be and is pushed. Without that (perceived and propagandized) pressure, IPv6 would long have become some exotic toy for some enthusiasts.
And yes, we all move from one standard to another one. But: Rarely it's people desiring or demanding a new standard. Usually it's governments or corporations or "committees" desiring and creating them. Sometimes for good reason, often, however, for doubtful or their own benefit.
Finally, IPv6 isn't just joy and richess (in terms of address space and new gadget features). It also has cost, very, very considerable cost. Software must be adapted or even rewritten (which also translates to some bugs being found but new ones being introduced), many related protocols must be adapted and/or retested (security, anyone?), new chips must be designed and of considerably higher complexity, thousands and thousands of organisations must do changes in their infrastructire, sometimes major ones (introducing errors and security risks along the way), etc, etc.
Take up is slow because many don't see any (to them) relevant or important progress but lots of problems, work, changes, trouble. The cost/profit ratio of IPv6 is and is perceived as unattractive.
Sure, most probably in the long run it will be force-fed and people will just have to eat it, period. But that's not a good way to make friends.
Just wait for the first IPv6 related security problem. Waves of anger will hit the IPv6 evangelists and enforcers. And those problems, and major ones at that, are bound to come (as with any highly complex technology).
Last but not least, without any desire to get political, but it's a factor whether we like it or not: The world is on a pretty unstoppable track towards a multipolar system. And there will be ugly questions why we had to bear many billions of costs just because the ex-hegemon usa kept an insanely improportionate large part of an important resource for itself. And why it were, of course, mainly usa corporations who profitted from the change to IPv6.
I can live with human weaknesses like mistakes made in engineering discussions. But I hate lies and I hate dirty money making schemes. And IPv6 is badly riddled by both.
@bsdguy: Just curious, so I can better understand your perspective: Where are you? Are you in the US?
No, I'm in Europe. And btw, we here in Europe are a little better but not really much (in terms of IP4 waste and insane allocations). Basically this comes down to 1% super-rich vs 99% people but here it's the version "A few western countries (usa + friends) have the large part of IP4 and the other 75% (or whatever) of the world must share the crumbs".
Pepsi is just an example but a good one. They have more IP4 than many large cities or even some countries. THAT's the problem. That's behind the "depletion".
And there are people who in all seriousness argue that it would burden Pepsi (and others) too much to switch their private networks to private IPs and that hence, we need IPv6 (which to switch to would, of course, be by far bigger a burden, but hey, it's IPv6, it's the messiah of internet, so we don't need to think rational ...)