North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

RE: PC Bozo's World bites again (CNN, too)

  • From: Doug Stanfield
  • Date: Fri May 29 10:25:44 1998

	The explanation lies in transiting circuits with high latencies
and/or packet loss, whether they're the last mile by dial up or frame
circuit backbone connections.  Larger packets are more prone to be
corrupted or discarded so sessions with smaller packets have a better
performance in comparison.  No excess retransmission and all that.
We've seen it in networks where the customer connection is high quality
but there were provider latency/loss problems.  Smaller MTU helped while
the frame relay problems persisted.  Once the provider network was up to
snuff there was no discernable performance difference between large or
small and I think most of our users went back to defaults.   

	> The improved-performance-with-low-MTU is a puzzle I'd
certainly like to
	> understand better.  It seems step 1 is to eliminate the
possibility that
	> the 1500 MTU is leading to fragmentation.

> 		Vern
	 
Doug Stanfield          			Oceanic Cable

Data Networking Manager        	200 Akamainui St.
[email protected]        		Mililani, HI  96789