“IPv6 is ready as a technology, but implementations aren’t.” At least that was my understanding of the presentations we had during the ICT Spring conference in Luxembourg last week. Several speakers came to talk about the technology. All are convinced about the need for IPv6.
IPv6 will be needed as the general vision is that we are going towards an “internet of things”. This means that not only traditional computers will be speaking IP, but also any kind of device : TV, mobile phones, domotics, sensors, cars, etc. The number of communicating devices will outgrow the number of inhabitants fairly soon. And all these devices will need an address.
The range of official IPv4 unique addresses will be consumed for September 2011. The shortage, which was already expected for somewhere end of the 90ies (yes, the nineties) was intercepted with Network Address Translation. This function allowed to have an official address on the outside and have thousands of “unofficial” addresses on the inside. Everybody has been using this. Issues can arise if your message travels across multiple of these NAT boxes. Network Address Translation will therefore not stand the pressure for additional addresses for a long time. An exponential growth in devices implicates an exponential growth in NAT tables.
A case study done by SpaceNet to support the popular german social network site “lokalisten.de” in their wish to move to IPv6, indicated that not all equipment manufacturers are really implementing the new standard. Their experience showed that the Operating Systems on the servers (Microsoft, Linux) were dual-stack (IPv4 and IPv6), and that the websites could handle it. But the load balancers placed in front of the servers could not handle IPv6 packets nor doing load balancing. Surprising, no, that a manufacturer of networking equipment doesn’t support the new protocol ? Another finding was that the database systems in use (don’t know the name) were not supporting the protocol neither. But as these were placed on the “inside”, and since the OSs are dual-stack, this could be solved easily.
All this points out that migrating to IPv6 will not be completely transparent on the inside. Load balancers, firewalls, databases, websites, … they all need to support it. Tests will need to be done to ensure a migration path from one to the other standard. I do not doubt that we will see more and more products coming out with implementations supporting the new protocol.
On the customer side, they made the website available in IPv4 and IPv6. 0.25% of the traffic was in IPv6 (3.5 million users), so even with a small percentage, it represent a fair amount of traffic. They have had zero complaints ! So on the user side it was completely transparent. And that is the proof of already years of development.
Most Comments