John E. Perry wrote:
Per Jessen wrote:
James Knott wrote:
Per Jessen wrote:
Adam Tauno Williams wrote:
]
* is faster for local transfers (gone are MTU/MRU & fragmentation issues)
How does IPv6 do that - isn't MTU dependent on the ethernet packet size and router support for the same?
When was the last time you saw anything that didn't support 1500 byte MTU?
This morning :-) I have an IPv4 setup which uses IPIP tunnels on the back of LVS, and those tunnels reduce my MTU to 1440.
Still, how does IPv6 avoid MTU/MRU & fragmentation issues? PMTU is surely still required.
...But isn't fragmentation a good thing? And isn't it good to be able to specify how much fragmentation to allow? How, otherwise, does one avoid a 50MB file transfer, for instance, locking up a 54Mbps wifi network for 10 seconds? Is provision made for such situations? If so, how does it differ from MTU? Or does it defer all packet-splitting to the lower levels? Are there any bottom-level protocols that don't split transfers into packets?
totally ignorant of ipv6, and not very knowledgable about _any_ networking,
John Perry
A large file, such as that 50 MB one you mentioned, is already split into small chunks by the transmitter, typically at 1500 bytes or so. The purpose of fragmentation was so that links with different size MTU could work together. So, for example, a 1500 byte packet would be split into 3 pieces, so that it could travel on a link with only a 576 byte MTU. Further fragmentation may take place if an even smaller MTU is encountered. All the pieces are then reassembled at the destination. With IPv6, path MTU discovery is used, which will determine the maximum MTU that can pass along the entire path, thus avoiding fragmentation along the way. The big problem with fragmentation is the work load it places on the routers along the path. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org