Author
|
Topic: Are we running out of IP addresses? (Read 1256 times) |
|
Walter Watts
Archon
Gender: 
Posts: 1571
Reputation: 8.89
Rate Walter Watts
Just when I thought I was out-they pull me back in
  
|
 |
Are we running out of IP addresses?
« on: 2008-04-16 17:42:41 » |
 |
This story appeared on Network World at
http://www.networkworld.com/news/2008/021308-ipv6-delay.html
Could IP address plan mean another IPv6 delay?
Proposal to allow IPv4 address trading could prolong Internet upgrade
By Carolyn Duffy Marsan , Network World , 02/13/2008
Internet policymakers are considering sweeping changes to the way they distribute IP addresses that could allow network operators to make money by transferring unused blocks of IPv4 address space to others in need. One result could be lessened incentive to move to IPv6 any time soon.
The American Registry for Internet Numbers (ARIN) is expected to post proposed changes to its IPv4 address space transfer policy on its Web site this week. ARIN is a nonprofit group in Chantilly, Va., that doles out IPv4 and IPv6 address space to ISPs operating in the United States, Canada and the Caribbean.
Under the proposal, ARIN would allow ISPs to transfer IPv4 address registrations, and ARIN would provide a list of IPv4 address blocks that are available for transfer.
Until now, IPv4 addresses have not been tradeable goods. When an organization was done using IPv4 address space, it was supposed to return it to one of five regional registries such as ARIN in North America. The only time ISPs can transfer IPv4 address space is when they are acquired.
ARIN's proposed changes are designed to help network operators cope when the Internet runs out of IPv4 address space, which is expected to occur in 2012.
"Industry demand for IPv4 addresses will not stop, but the current supply channel, namely the unallocated IPv4 address pool will have run out," says Geoff Huston, an expert on IPv4 address depletion and chief scientist at the Asia Pacific Network Information Centre, the Australian counterpart to ARIN. "So, as with any other commodity out there, trading and pricing gets included into the distribution function."
IPv4 is the Internet's main communications protocol. IPv4 uses 32-bit addresses and can support around 4 billion IP addresses. IPv6 is a long-anticipated upgrade to IPv4. IPv6 uses a 128-bit addressing scheme and can support so many billions of IP addresses that the number is too big for most non-techies to understand. (IPv6 supports 2 to the 128th power of IP addresses.)
The IETF designed IPv6 in the mid-1990s to expand the available IP address space. However, few ISPs or enterprises have upgraded to IPv6.
The issue of IPv4 address depletion has received a great deal of attention in the last few months. Experts say more than 80% of IPv4 addresses have been distributed.
Huston says it is too late for the Internet to avoid creating a way for ISPs to transfer their titles for IPv4 address space because the Internet will run out of the available pool of IPv4 addresses before everyone transitions to IPv6.
"We now need to talk openly in our policy development process about transfers, trading and mechanisms that will allow the Internet to continue to function as smoothly and as reliably as possible in the coming few years," Huston says.
APNIC and the European registry RIPE are considering similar changes to their IPv4 address transfer policies as ARIN's proposal. These changes are controversial and may not be approved.
Many unanswered questions surround IPv4 address trading:
• Will it create a financial market for IPv4 address space? (Read what the Internet Assigned Numbers Authority has to say about the potential for an IP address black market.)
• Will it delay the transition to IPv6 because more IPv4 addresses will come available?
• Will IPv4 address transfers swamp the Internet's core routers with too many routing table announcements from ISPs?
"We don't know whether some of the side effects of such a policy makes sense for the Internet," admits John Curran, chairman of the ARIN Board of Trustees. Curran is chief technology and operating officer at ServerVault, a Dulles, Va., managed security services provider.
If IPv4 address trading is permitted, the likely beneficiaries are the U.S. federal agencies, universities and companies that received massive blocks of IPv4 address space at the dawn of the Internet. Back then, no one realized that IPv4 addresses would become a precious commodity. So they didn't assign IPv4 addresses efficiently across their wiring closets, buildings and campuses.
Until now, these organizations have lacked a financial incentive to renumber their networks to free up IPv4 addresses. It is rare for an organization to return extra IPv4 addresses. Notably, Stanford returned more than 16 million IPv4 addresses in 2000.
With an IPv4 address shortage looming, policymakers are stepping up their efforts to recover unused IPv4 address space. This week, the Internet Corporation for Assigned Names and Numbers (ICANN) announced that it had recovered 16 million IPv4 addresses from Net-14, which was originally used to connect older packet data networks.
ARIN's proposed IPv4 transfer policy would provide an economic incentive for organizations to free up IPv4 addresses.
"Hypothetically, a large company with excess IPv4 address space could get compensated for the work of freeing up that space," Curran says.
ARIN's proposal wouldn't allow speculation on IP addresses as has occurred with domain names because it requires IPv4 address space that gets transferred to be used.
No one knows if sizeable profits could be made from transferring excess IPv4 address space.
"Now we're telling people that [returning unused IPv4 address space] is the right thing to do without compensation," Curran says. "When you set up a process where an organization can be compensated so it can free up address space that others might not have, it's very hard to say how that system will actually behave."
The U.S. Defense Department, for example, is sitting on a mother lode of IPv4 addresses. Could this become a sale-able asset for the Department of Defense, akin to a wireless spectrum auction? Experts say that scenario is unlikely.
"It's fairly difficult to imagine circumstances where the receipts for such a transfer policy would be so large as to incent someone who was using the address space to actually stop using it," Curran says.
Experts agree that allowing the transfer of IPv4 addresses would likely delay the transition to IPv6 by several more years.
"One of the forecasts that's most common says that if the un-advertised IPv4 address space were somehow put back into use that could push out the date of IPv4 address depletion by another five or six years," Curran says. "Yes, I think allowing IPv4 address transfers could move back the date for IPv6, but I don't know to what extent. It could be months, or it could be a handful of years."
Not everyone thinks IPv4 address trading will delay IPv6 deployment. "I doubt it would make much difference," says Scott Bradner, a data networking expert at Harvard University, ARIN trustee and a Network World columnist. "It might even speed it up when companies who can switch [to IPv6] have an additional reason to switch in that they could sell off their old [IPv4] space."
Bradner is skeptical that government agencies, universities or corporations can make much money by selling excess IPv4 address space. "It could easily cost an organization much more to renumber out-of-range addresses than they would get from selling the space they might free up," he explains.
Most U.S. network managers have not yet begun migrating to IPv6. At issue is how these network managers will continue to grow their networks once the unallocated pool of IPv4 addresses runs out. IPv4 address trading may solve that problem, experts say.
"ARIN allocated some 53 million IPv4 addresses in 2007. Demand for IPv4 addresses appears to be continuing," Huston says. "If we are still using IPv4 in a couple of years time, even as part of a dual-stack transition to IPv6, then network managers of growing networks will have a continuing need for more IPv4 addresses…Consumption rates on the unallocated address pool raise the question of how that future demand for IPv4 addresses will be met."
ARIN's Advisory Council, a 15-member group of network engineers employed by ISPs and universities, drafted the proposal to change the IPv4 address transfer policy.
Network managers are encouraged to comment on the proposed changes. The changes will be up for discussion at ARIN's next meeting, which will be held in Denver from April 6-9.
All contents copyright 1995-2008 Network World, Inc.
http://www.networkworld.com
-----------------------------------------------------------------
On a related note, here is a great article (and site)
covering IPv6 and related issues (NAT in this particular article):
http://www.ipv6.com/articles/nat/NAT-In-Depth.htm
Walter
|
Walter Watts
Tulsa Network Solutions, Inc.
No one gets to see the Wizard! Not nobody! Not no how!
|
|
|
Hermit
Archon
Posts: 4287
Reputation: 8.94
Rate Hermit
Prime example of a practically perfect person
|
|
Re:Are we running out of IP addresses?
« Reply #1 on: 2008-04-17 03:04:58 » |
|
We haven't, are not and will not run out of IP addresses. Only a tiny fraction of those allocated are used.
What we have run out of is the ability of routers to determine where to send packages, because routing decisions depend on being able to determine - fast if you don't want massive performance impacts - where to forward messages to get them to their eventual destination. Routers can work this out because addresses are issued in blocks which allows address aggregation. You might think of address aggregation in terms of mail. A router far from the destination doesn't need to know the name or PO Box of the recipient, only the target country (and maybe the target city)).
So too, routers far from the destination don't look at the whole address, because addresses are allocated in blocks, they only need to look to see which port is best (best in terms of fewer hops, more bandwidth, less congestion or whatever routing algorithm or algorithms is used) suited to get the package closer to the block of addresses the destination is associated with. When this was managed by the sub-allocation of blocks and the address space scarcely used, it was relatively easy to keep track of the top level blocks; and as the packets got closer to the destination, the routers would have more detailed information closer to hand till finally the packet would make it to the appropriate network holding the destination.
The trouble is that this very scheme meant that we handed out addresses in blocks that matched powers of two. It began with 3 classes of (conventional addresses, Class D (broadcasting) and E (experimental) are not used the same way) IP addresses, Class C (24 bit network address, 8 bit host address), class B (24 bit network address, 8 bit host address) and Class A (8 bit network address and 24 bit host address). Obviously most networks have far fewer hosts than the number of addresses allocated, and so, by 1998 we ran out of blocks of addresses to allocate. We solved that problem by inventing CIDR (which allows any power of 2 to be used as an address boundary, allowing, e.g. a /30 network (with a mask of 255.255.255.252) allowing a 2 host network, as well as private addressing together with ways of transmitting and receiving data between the private addresses and the Internet; allowing a few private addresses to be reused all over the world.
The right answer, we knew back then already, was to switch to IP v6, Where every grain of sand on the planet could receive its own network address and still have addresses free. Unfortunately, IP 6 addresses were not handed out rationally in big blocks, they were closely hoarded at the upper levels and ISPs have tended (and still tend) to charge users to allocate them IP 6 addresses, where they would provide IP 4 addresses free of charge. Exactly the wrong way around. This has meant that few users have designed their networks for IP 6, and ISPs hating to redo perfectly good systems and retrain staff who can't even deal with IP4 addressing issues, are reluctant to introduce such fundamental change. So between 1998 and today we have been reusing bits of address space and routing smaller and blocks on the Internet. This has had huge implications for backbone routers which have needed ever larger memory space, but has not impacted ISPs to any significant extent.
I see the proposals Marsan has written about as being merely an extension of the actions taken in 1998. They will impose additional enormous costs on backbone providers who will need much larger router memories as well as far more bandwidth for routing even with aggressive use of route aggregation and modern routing algorithms, will provide much less breathing space and will reduce network performance. In my opinion, the correct answer remains that which existed in 1998, which is to encourage and if needed, mandate moving to IP 6. A better interim strategy than that above would be for the backbone providers to begin charging differential connection fees for the kind of networks run by people connecting to them with a large surcharge for pure IP4, a medium surcharge for transitional networking and no surcharge for IP6 alone. This would match the costs incurred and the revenue generated, and would strongly encourage a transition. This wouldn't be difficult to do, as the backbone providers already allocate the blocks to those further down the line - and could easily allocate big blocks of IP6 addresses. And it would be nice to have enough IP addresses to make every machine connected to the Internet visible to the Internet again. Although of course, so long as people keep loading Microsoft products onto otherwise innocent hardware, that would probably not be a universally good idea.
Kind Regards
Hermit
|
With or without religion, you would have good people doing good things and evil people doing evil things. But for good people to do evil things, that takes religion. - Steven Weinberg, 1999
|
|
|
Fritz
Archon
Gender:
Posts: 1746
Reputation: 8.84
Rate Fritz
|
|
Re:Are we running out of IP addresses?
« Reply #2 on: 2008-05-17 01:26:48 » |
|
More IPv6 buzz .... I still liked Decnet better :-)
Cheers
Fritz
http://www.theregister.co.uk/2008/05/16/oecd_ipv6_report/
World economy group gives IPv6 big push
Warns about depletion of IPv4 addresses
By Kelly Fiveash
Published Friday 16th May 2008 16:09 GMT
The Organisation for Economic Cooperation and Development (OECD) sounded an alarm bell yesterday over the rapid depletion of IPv4 internet addresses and gave the IPv6 protocol another push.
In the new report titled The Future of the Internet Economy, which has been published ahead of the group’s ministerial meeting in Seoul next month, the OECD supports the widely-held view that the currently-used version of the Internet Protocol (IP) will run out of addresses in 2011.
It observed that “beyond building IPv6 skills and applications within governmental bodies, public procurement mandates also lead to a virtuous cycle of adoption by instigating the development of skills within technology partners".
Network Address Translation (NAT), which makes it possible for several systems to share a single IPv4 address, is already widely used.
But it’s a stop gap system, viewed by some observers as an imperfect and costly work-around. The report claimed that enterprise and application vendors spend as much as 30 per cent of IT-related expenditures on the system’s sub-par communication protocol tweaks.
The report echoes internet search giant Google’s call earlier this week for the wider adoption of IPv6 as a long-term solution to what is becoming a growing concern for the tech industry.
“While technologies such as Network Address Translation can offer temporary respite," it said, "they complicate the internet's architecture, pose barriers to the development of new applications, and run contrary to network openness principles.”
Google also took the opportunity to point out that its search facility was now available over IPv6. It even gave Microsoft's unloved OS a shout-out.
“With current operating systems such as Windows Vista, Mac OS X, and Linux providing high-quality support for IPv6, we hope it's only a matter of time before IPv6 is widely deployed,” it said.
The full report can be read here here (pdf). ®
|
Where there is the necessary technical skill to move mountains, there is no need for the faith that moves mountains -anon-
|
|
|
|
Logged
