Thursday, December 1, 2011

Why Move To IPv6?

       I did a post last week titled Are You Ready for IPv6? where I share some of my thoughts on IPv6 and the allocation of the last IPv4 /8 blocks. Since then, I've done some more reading and found a couple of very useful podcasts by Greg Ferro and Ethan Banks. It is where I first heard about v6RD.

There are real reasons to move to IPv6. I recall a few statistics and examples from years ago that illustrate the need for IPv6. Addressing every school in China would have exhausted the available IPv4 address space even a few years ago when many more IPv4 addresses were available. The other example is addressing all the power meters in the country. Think about all the power meters you see around and the effort that goes into reading them on a regular basis. Addressing each meter and building a way to network them back to the power provider allows them to be read with fully automated mechanisms. That volume of addresses that was required was also bigger than the then-available IPv4 address space (and perhaps even a big chunk of all the IPv4 address space, regardless of whether it is used or not). So we definitely need more address space. And some of the functions within IPv6 are useful, such as auto-configuration, which would be very beneficial for power meter addressing.
  • Training the network staff to handle IPv6 configuration and troubleshooting.
  • Any application that uses IP addresses internally or that sends IP addresses in messages to other systems will need to be modified.
  • pplication developers will need to modify and validate applications to use DNS to translate system names into IPv6 addresses. Some applications are very expensive to modify. I know of several health care applications and products that use hard-coded IP addresses (no DNS) and that the vendor requires payment to modify embedded addresses. Because many of these products are certified by the Food and Drug Administration, they cannot be modified without going through another validation and acceptance process. I don't see this happening anytime soon.
  • Dual-stack support (or a similar mechanism) will be required by the e-commerce sites because many customers will be on legacy IPv4 networks while other customers will have transitioned to IPv6.
  • Firewalls and ACLs will need to be updated to perform equivalent functions for both IPv4 and IPv6. Keeping changes to firewall rules and ACLs in sync will be problematic at best.
  • Network management systems will need to handle IPv6. Address fields will need to be much larger. The NMS databases will grow in size and NMS developers will need to spend time looking at how they display device and interface information. And how is the NMS supposed to display the information about an interface that is configured for both IPv4 and IPv6? Maybe IPv6 will force the industry to start using logical names instead of addresses.
  • What does IPAM in IPv6 look like? Each subnet is a /64, on which there are 2^^64 host addresses. Displaying an IP address map of the entire address space doesn't make sense. Perhaps it should display the locally-assigned part of the address (exclude the site prefix and the host part of the address) and show the number of systems that exist in the subnet. IPAM will still be needed to help track which subnets have been allocated and where they are allocated. If you want to track end stations, the NMS will need to query the routers and switches to find the end station addresses or you'll need to use DHCP for IPv6.
  • Firewall rules and ACL entries need to be replicated into IPv6, with certain exceptions for ICMP that IPv6 uses for neighbor discovery.
In summary, I see a lot of costs and not much economic benefit to offset those costs. However, I'm looking forward to exhaustion of IPv4 space. It is going to create a whole new line of consulting and network management work as companies start to figure out what to do about it. You can think of it as the Y2K event of this decade.

The points I make above are why organizations won't move to IPv6. It is quite possible that I've overlooked some reasons why organizations should aggressively move to IPv6.  The only incentive that I can identify is for organizations to be internationally competitive. Any information that you may have regarding economic incentives to implement IPv6 are greatly appreciated. Please post a comment if you know something that I have overlooked.

For some additional reading, check out the following links:

Thank you to

One of the things that I've not seen addressed is the economic cost of IPv6, which is the factor that has the greatest impact on how quickly IPv6 is adopted. New networks outside North America have an incentive to use IPv6: they don't have to redo their network at a later date. So we are seeing IPv6 adoption at a much faster rate internationally than in North America. It makes sense to save time and money by starting with an IPv6-capable infrastructure. It also allows those countries to develop their expertise in v6, making them more competitive with respect to other countries.

But in North America, what incentives exist for companies to start using IPv6? The customer base is primarily IPv4, so supportinging customer contact is not a good argument. Except for the power meter example above, there is enough address space. NAT has slowed the rate of address consumption, making it less of an incentive. I don't see any financial incentives outside that of being able to communicate with international customers who are using IPv6 and are not connected to an ISP that translates to IPv4.

What about the costs of IPv6? I see a lot of costs:


Post a Comment