Will IPv6 networks be ready to handle government's needs?

The next version of Internet Protocols is essential to the Defense Department's net-centric vision, and the government has committed to having its networks running IPv6 by 2008.

Will those networks provide the performance needed to connect a multitude of new devices with real-time communications to provide a level of situational awareness not available now? Nobody knows for sure. But for the time being, you cannot expect IPv4 speeds on an IPv6 network.

Latency grows and performance decreases for many network devices processing IPv6 packets, said Bill Kine, a product manager for Spirent Communications Inc. of Calabasas, Calif.

'That is a reality of having to process the 128-bit addresses,' Kine said at last month's U.S. IPv6 Summit. 'We're asking the devices to do a little bit more.'

Although most networking hardware and software, and some applications, are IPv6-capable today, they are designed for IPv4 as the primary protocol. Many of the new protocol's benefits are provided by its larger packet headers and addresses, which provide more information but also require more work to process.

Putting IPv6 packets on an IPv4 network is like putting a bunch of Hummers on a highway designed for Volkswagen Beetles. No matter how powerful the new vehicles are, they aren't likely to get anywhere too quickly.

This situation is likely to change as IPv6 becomes the protocol of choice and products are designed to its specifications. But for now, we know little about how IPv6 packets really work on our networks.

'We're making a lot of progress in IPv6 testing, but we can't pat ourselves on the back yet,' Kine said.

Spirent, a manufacturer of testing equipment, has partnered with the consulting firm V6 Transition to establish an IPv6 testing center in Northern Virginia that will help answer these questions. V6 Transition is a subsidiary of Innofone Inc. of Santa Monica, Calif.

Although testing for conformance with IPv6 standards has been available for five years, functional testing of equipment is just getting started.

So far we know that networks can forward IPv6 packets and that end devices can cooperate with IPv6, although most implementations have been over small, in-house networks.

What we don't know

But we do not know much about how well equipment actually performs with IPv6, or when running a combination of versions 4 and 6. What is the recovery time for failures? What is the speed of routing?

Scalability is an issue. The 128-bit IPv6 addresses are four times as large as the current 32-bit addresses used in IPv4. Will they require four times as much memory in a router?

Kine said that in tests, a small edge router that could hold 50,000 IPv4 addresses could accommodate only about 15,000 IPv6 addresses. He has not tested it using both at the same time.

Quality of service is another issue. Performance parameters of IPv4 devices are well known in most circumstances so that service levels can be guaranteed.

'Version 6 performance is all over the chart,' Kine said. 'We don't know how to guarantee service levels for it.'

These questions eventually will be answered and improvements made, but in the meantime, IPv6 is not likely to be a quick fix for performance needs.

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above