🚀 go-pugleaf

rmwise@mcigate.apdev.mci.com (Bob_Wise) writes:
>
>I've been following this packet-fragmentation-when-using-NFS-over-FDDI
>thread with some interest.
>
>I've got two IBM RS/6000 950's which are shortly to become 970's, plus
>I've got another 2 970's coming.  We do a lot of NFS over ethernet,
>and I am about to move that NFS traffic over to a small FDDI ring that
>only has the 970 servers on it.
>
>Is the fragmentation problem present with this scenario, or is
>fragmentation only a problem with bridges and routers?
>
>Do I need to consider adjusting packet sizes?
>
>Thanks,
>
>-Bob
>
>
>
>-- 
>---------------------------------------------------------------------------
>| Bob Wise          | INET:622-1322 | MCIMail:rmwise | Pager:719-577-1928 |
>| Unix Consultant   |-----------------------------------------------------|
>| Consultant to MCI | Phone:719-592-1322 | Internet:rmwise@apdev.mci.com  |

Bob, et.al.,
   If you have the opportunity to to design a net from the ground up
I'd suggest you look at using routing *IF* you will be spanning
multiple media (fddi, ethernet, tokenring). When you start trying
to do bridging between different types of media all kinds of nasty
issues raise their head. These issues are effectively dealt with
by routing. In case you are wondering what types of "nasty issues"
I'm refering to here is a short list off the top of my head I'm sure
others can add to it.

 o MTU mismatches (fddi vs eth)
 o Some ARP implementations use canonical address within the data
   others use non-canonical address representation within the data field.
 o Source Route to non SR media (token ring to ethernet or fddi)

-brad-

--
v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v
Brad Turner | 5400 Bayfront Plaza  | Marketing Engineer   | (408) 764-5261
3Com Corp.  | Santa Clara CA, 95052| mbt@NSD.3Com.Com     | (408) 764-5002 fax