On Mon, 2 Feb 1998, Jukka Marin wrote:

> BTW, what's the basic idea of DF, anyway?  The fragmented packets will be
> reassembled at the receiving end, so why not allow fragmenting and let the
> routers do what they have to? :-)  (A stupid question probably, but it had
> to be asked.. ;)  Sure, it may affect performance with high packet losses,
> but..

Yes, the packet could be reassembled, but that takes time and buffer
space.  So, lots of systems do Path MTU Discovery by sending packets
with the Don't Fragment bit set.  If a router along the way would
otherwise have fragmented the packet, it instead discards the packet and
returns an ICMP Need-to-Fragment to the sender instead.  So the sender
can then try a smaller packet size, eventually learning the largest
packet size that can be sent end-to-end without fragmentation.

-----------------------------------------------------------------------------
| Paul Goyette       | Public Key fingerprint:   | E-mail addresses:        |
| Network Consultant |  0E 40 D2 FC 2A 13 74 A0  |  paul@whooppee.com       |
| and kernel hacker  |  E4 69 D5 BE 65 E4 56 C6  |  paul.goyette@ascend.com |
-----------------------------------------------------------------------------