| previous 3/7 next |
![]() |
| Vinton Cerf, senior vice president of Technology Strategy, MCI |
Throughout the history of the Internet, transparency has been a key requirement. But the Internet is spreading to undeveloped world. Some laundromats are offering Internet access to customers. We are beginning to see more Internet-enabled devices, such as net-connected digital picture frames.
IPv6 was developed through discussions sparked by the concerns in early 1990s that IPv4 address space would be consumed within a decade or so, although here is a more recent prediction that IPv4 address will last till 2020s.
Then, what are the real issues with IPv6?
Mass of software, from core to the edge, need upgraded. This is a minimum requirement for successful transition to IPv6.
There is a debate as to whether connected nodes should be single-stack or dual-stack. Single-stack leads to internetworking issues. Each node can only talk with nodes with matching protocol. Single-stack servers may have to be accessed through protocol translation. If NAT boxes persist, they make things more complicated. If dual-stack is preferred, then would that reduce the motivation of servers to adopt IPv6?
IPv6 support is necessary for various network service software, such as router, network management, DNS resolvers and servers, registry/WHO IS databases. ISP provisioning software and back office software needs to support the two protocols.
Should NAT boxes really go away? Some may say that NAT is necessary for mapping between IPv4 and IPv6.
IPv6 uses server-centric mechanisms for name resolution as IPv4 does. But new rendezvous mechanisms may become more useful than DNS in discovering peers.
In security, IPsec key distribution needs worked on. And PKI has worked for intra-organizational use, but less well for inter-organizational and public use. IPv6 nodes can have multiple addresses. Equipment service companies may assign distinct address for their products and provide necessary services with separate routing paths. Many tunneling technologies are available, but each has is own disadvantages. Manual tunneling lacks scalability.
Analysis and documentation would be helpful to guide IPv6 development requirements. Application drivers will be the key to providing ISP incentives to deploy IPv6. IPv4 is not likely to fully go away. But IPv6 is the only clean option to get more address space.
| previous 3/7 next |
この記事のトラックバックURL
http://www.ipv6style.jp/trackback/394



