x
Optical/IP

Mobile Access Drives IPv6

Recent software upgrade announcements from Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (Nasdaq: JNPR) reveal the killer application that will finally bring IPv6 (IP version 6) into the mainstream: mobile Internet usage. And from the looks of things, Asia is leading the pack in deployments.

Today, Cisco announced the availability of Phase II of its IPv6 rollout in Cisco IOS software (see Cisco Upgrades Gear). These features build on Cisco's previous release in May 2001 and provide more integration between the current IPv4 and IPv6. Last week, Juniper announced support for IPv6 across all of its routers (see Juniper Unveils IPv6 Routers). Both companies have cited mobile usage and growing deployments in Asia as the drivers.

Cisco and Juniper are the latest companies to announce support and enhancements for IPv6. Foundry Networks Inc. (Nasdaq: FDRY) and NEC Corp. (Nasdaq: NIPNY) both announced support back in October (see Foundry Routers Get IP Infusion and NEC Launches 'BlueFire'). Riverstone Networks Inc. (Nasdaq: RSTN) says it's on its roadmap for next year.

The fact that equipment vendors are starting to roll out products supporting IPv6 signifies that customer demand is reaching critical mass.

“I think the decision of these vendors to offer IPv6 is driven by customer demand,” says David Newman, president of Network Test Inc. “There is a lot of next-generation wireless deployment in Asia right now, and if there are enough customers asking for IPv6 the vendors will provide it. They are technology agnostic; they just go where the money is.”

IPv6 is in demand because it provides more addressing space, security enhancements, and network management. IPv4 is able to accommodate about four billion addresses, while IPv6 could potentially support billions more, according to a Cisco spokesperson.

While there are currently enough IP addresses, the advent of new Internet-enabled devices like Web phones and Internet-ready, hand-held devices will cause the current scheme to eventually run out. According to analyst firms like IDC, over the next three years the number of mobile Internet users could jump as high as 400 million.

Asia, which is exploding with Internet mobile users, has become a hotbed for early adoption of IPv6. Mobile Internet usage in the region is estimated to be up 52 percent in 2000 from 1999, according to Gartner/Dataquest. By 2003, Gartner estimates, more than one billion users will be using cell phones worldwide.

“We are currently beta testing IPv6 in 20 beta sites across all geographies,” says Kevin Dillon, director of product marketing for Juniper. “But demand seems to be greatest in Asia right now, where there is a lot of mobile Internet usage.”

While Asia seems to be leading the pack in deployments, Europe isn’t far behind, he adds. But in its early phases of deployment, IPv6 is mostly being deployed in research networks like SURFnet5's (see Cisco Powers IPv6 in Europe). France Telecom SA Research and Development and GIP Renater, in Europe, and 6TAP, in the U.S., are using Juniper’s IPv6 technology.

But IPv6 is not without its problems. Because the address field of IPv6 is so much larger than IPv4, it will result in extremely large routing tables that could be difficult to manage, says Network Test’s Newman. But Juniper’s Dillon says current routing tables have about 100,000 routes, and Juniper’s tables can handle three to five times that amount.

”I think that issue may prove not to be as big a problem as everyone makes it out to be,” says Dillon.

The other issue is the integration of IPv6 into the existing IPv4 infrastructure. Cisco’s Phase II and Juniper’s implementation both use Multiprotocol Label Switching (MPLS) tunneling to transport IPv6 traffic over existing IPv4 networks. But critics say this solution won’t scale for large deployments. Newman says tunneling adds complexity and overhead to the network, which can degrade network performance.

“Regardless of technology -- if it's IPv6, IPv4, or tin cans and piano wire -- if it doesn’t scale and provide a 10-times improvement in cost, it won't go anywhere,” says Newman.

Juniper says that its IPv6 is built into its latest ASICs. Dillon asserts that performance is not affected by the use of IPv6. He also says that both IPv4 and IPv6 will ship in the current software, allowing customers the ability to use both.

According to Cisco, its Phase II version integrates IPv6 with the IPv4-based Internet by providing protocol translation between the newer version and the existing protocol. There is also still more to come from Cisco. Phase III of its IPv6 rollout, which will roll out next year, will include native support for the OSPF routing protocol for large-scale deployments and hardware-accelerated IPv6 forwarding on the Cisco 12000 Series Internet Router.

Juniper’s IPv6 is now available on all of its routers: the M5, M10, M20, M40, and M160. Cisco's IPv6 Phase II rollout is only available now for limited deployment, with wider availability scheduled in the first half of 2002. Platforms supported at this time include Cisco's 800, 1400, 1600, 1700, 2500, 2600, 3600, 7100, 7200, and 7500 Series Routers and its 12000 Series Internet Routers.

Support for Catalyst 6500, an enterprise box, and Cisco 7600 Series edge routers, is planned to be available with a future release of Cisco IOS Software.

While most people agree that IPv6 won’t overtake IPv4 for some time, it’s fast becoming a necessary feature in closing deals, especially in Asia and Europe.

“If you’re going to compete for business for mobile and gaming networks,” says Dillon. “IPv6 is table stakes.”

— Marguerite Reardon, Senior Editor, Light Reading
http://www.lightreading.com
mjtate 12/4/2012 | 7:29:43 PM
re: Mobile Access Drives IPv6 I think it is fair to say that IPv6 address allocation / structure methodology hasnt resolved the problem of multi-homing between providers.

CIDR provided some of these benefits but as as seen in the "classic" Internet the cause of increasing address table size is Enterprise Multi-Homing between providers.

I refer you to the Geoff Houston presentation made in the 49th IETF San Diego meeting (2000)

There is still much more work to be done before IPv6 can be seen as Nirvana and multi-homing is one of the key requirements for it.

Best wishes
veemee 12/4/2012 | 7:29:43 PM
re: Mobile Access Drives IPv6 Also the heirarchical addressing in IPv6, will certainly reduce the size of the default free routing table.
veemee 12/4/2012 | 7:29:43 PM
re: Mobile Access Drives IPv6 IPv6 has still not started to be used in live networks(not the r&D networks). Are all vendors scrambling to put the fancy feature(cuz someone else has it), though it isnt going to be used in the near future.

When are the first deployments on the internet expected? How long will NAT boxes hold on?
rainbowarrior 12/4/2012 | 7:29:42 PM
re: Mobile Access Drives IPv6
I'm really surprised that V6 is getting any marketing traction at all at this point given its well-known technical problems.

V6 was orignally proposed to introduce a raft of benefits to IP including security, automatic addressing, flow awareness, extensibility and expanded addressing. Each of these benefits was adapted to V4 ( IPSec, DHCP, Diff-Serv, MPLS ) so that the only advantage over V4 left is V6's unnecessarily large addressing space.

The huge addressing space causes routing table expansion in the core. Yes- agreggation can mitigate this problem - but multi-homing cancels the benefits of aggregation. Routing stablity in the core of the V4 Internet is challenging enough. Exponential expansion of routing tables would make it impossible.

Add to that simple problem countless more basic ones such as the fact that simple V6 provisioning is a nightmare because the addresses are so incomprehensible to humans, lack of good DNS support, routing protocols that at best have been untested in large-scale networks and you see why there are no large V6 deployments.

And as far as wireless being the killer app- wireless devices are using V6 address as MAC addresses. They are by-in-large used for single hop communication and mapped directly into VCs rather than routed.

The whole concept of expanded addressing is just propping up the antiquated concept that every device on the internet needs at least one permanent layer 3 address. It's solving the wrong problem.

It's quite ironic that so much of the useful subset of the work in the IETF lately has been trying to get around the limitation of a universally accessible host-addresss-per-device. IP VPNs, IPSEC/IKE, NAT and MPLS could all be viewed as examples of technologies supporting a temporary session rather than a permenant, universal, endsystem addressing structure.

After all, most internet traffic today ( such as web-browsing and e-mail ) are based on a session-oriented protocol ( TCP ). If we worked toward supporting a session-based protocols that allowed addresses to be time-shared and/or domain specific, than trying to upgrade the entire internet to support a massively overprovisioned addressing structure.

rriicc 12/4/2012 | 7:29:36 PM
re: Mobile Access Drives IPv6 Security-wise NAT/NAPT is pretty good (yes, it's a side effect). All these mobile terminals, phones, PDA to have public IPv6 address might be a security and resource management nightmare. The radio spectrum is much more restricted compared to the wireline, where the spam, DOS etc. attacks can be launch from. It can be very hard to filter out all bad traffic in the gateway to the mobile access network. And filters in the terminals are too late (radio resources are already consumed).
An alternative to IPv6 is to use application based "addressing" and have the mobiles notified over broadcast radio channel in the cell for incoming "calls". Than they the mobile can estabilish a session thru the NAPT gateway.
The applications have to be built NAT transparent as they should according the OSI model.. Most application works across NAT already.

So, IPv6 is not needed... even if there where public IP address for all user one wouldn't use it in the mobile Internet :-)
greybeard 12/4/2012 | 7:28:36 PM
re: Mobile Access Drives IPv6 V6 was orignally proposed to introduce a raft of benefits to IP including security, automatic addressing, flow awareness, extensibility and expanded addressing. Each of these benefits was adapted to V4 ( IPSec, DHCP, Diff-Serv, MPLS ) so that the only advantage over V4 left is V6's unnecessarily large addressing space.
greybeard 12/4/2012 | 7:28:35 PM
re: Mobile Access Drives IPv6 Apologize for the previous null response. My system got carried away.

> ...so that the only advantage over V4 left is
> V6's unnecessarily large addressing space.

Yes, the large address space is the only point.
However, having a space this large can be very
useful, if only folks can use their heads to
figure out how to use it.

> Routing stablity in the core of the V4 Internet
> is challenging enough. Exponential expansion of
> routing tables would make it impossible.

You need to distinguish between number of routes,
versus the size of each route.

25 years ago routers did just fine with 32 bit
addresses. Processors and silicon have become a
lot more than 4 times faster since then, and will
have no problem with 128 bit addresses.

The number of routes should grow by less than a
factor of two, since the IPv4 space has to deal
with "legacy" address assignments, and IPv6
doesn't. A factor of two is not much to worry
about on this scale (it is factors of ten or a
hundred which we should worry about). Of course,
the doubling in numbers of routes comes from
having both v4 and v6 routes.

> V6 provisioning is a nightmare because the
> addresses are so incomprehensible to humans

stateless address auto-configuration.

> lack of good DNS support,

Its the same DNS

> routing protocols that at best have been
> untested in large-scale networks

BGP and IS-IS are essentially unchanged. Both
already can deal with multiple address families,
and IPv6 just adds one more. OSPF has more
changes, but is still pretty old and well
known solid technology.


If you guys really want to pick on something
real, look at the transition schemes. The
protocol is solid. Products are coming. It is
the transition where there will be surprises.

HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE