Path: utzoo!utgpu!attcan!uunet!husc6!rutgers!apple!bionet!agate!saturn!eshop
From: eshop@saturn.ucsc.edu (Jim Warner)
Newsgroups: comp.dcom.lans
Subject: Re: Errors on a LAN
Message-ID: <5379@saturn.ucsc.edu>
Date: 6 Nov 88 10:04:03 GMT
References: <1052@esunix.UUCP>
Reply-To: eshop@saturn.ucsc.edu (Jim Warner)
Organization: University of California, Santa Cruz
Lines: 34

In article <1052@esunix.UUCP> dirish@esunix.UUCP (Dudly Irish) writes:
>We have a large LAN at our site partitioned into 9 LANs by DEC bridges.
...
>And we see on the order of %0.005 packets have one of these errors.
...
>So would the people out there will similar sized networks please post your
>experiences?

You don't say how large the segments between the bridges are.  That's
significant because these errors aren't propagated through LanBridges.
Your error rates don't seem at all out of line.  You should be aware
that when the error rate is this low that the point of attachement
of the monitor on the LAN will probably affect the results.  Looking
at a dual interface Sun server connected to two ethernets, I get:

total       packets
packets     with
rcvd        errors

 903607      33       Single segment sun-only net
2474071      13       Multi-segment four building backbone

The multi-segment backbone uses repeaters which *DO* propagate
damaged packets.

I start getting interested when the bad packet rate climbs to
about 0.05 %.  I get concerned when it climbs above 0.2 %.

I could make these error rates much lower by dumping lots of
money into improved network hardware.  But I don't see that
it would be worth it.

jim warner
U.C. Santa Cruz