Why do you think IPv4 does fragment reassembly at the end rather than at the next router?
Fragment reassembly is done at the end host always in order to free routers from having to do this task. Routers need to be efficient in forwarding, so any task which is not directly related to forwarding and can be done somewhere else is a gain in performance.
In IPv6 it goes even one step further. With IPv6, routers do not even fragment anymore. If fragmentation needs to be done, it is done at the sending host.
For more information:
Dig Deeper on Network Infrastructure
Related Q&A from Silvia Hagen
An expert explains why the adoption rate of IPv6 isn't as slow as people think and why it is important to create a dual-stack environment during the ... Continue Reading
Expert Silvia Hagen explains what v6LoWPAN does and how it’s used. Continue Reading
Expert Silvia Hagen explains what ICMPv6 is, and how it differs from ICMPv4. Continue Reading