Everhart, Glenn
From: carl@gerg.tamu.edu
Sent: Wednesday, February 10, 1999 12:41 AM
To: Info-VAX@Mvb.Saic.Com
Subject: Re: Vax Cluster Copy File Performance
"Richard B. Gilbert" <76702.1567@compuserve.com> writes...
} My reference material is not where I can lay hands on it
}easily but I seem to recall a distance limit on Thin-net that is less than
}100 meters. I think it was something like 185 feet! Even if your cable
}is too long, it might work. We won't know until you tell us what's
}happening!
Right number, wrong units.
"Quickly, the rules for ThinWire (10base2) Ethernet are:
- per-segment:
- at most 30 stations on a segment (including repeaters if any)
- max segment length: 185 meters
- at least 0.5m between stations (<=from memory - could be a little off)
- per collision domain (i.e. repeatered Ethernet):
- at most three coax segments between any two stations
- therefore, at most two repeaters in the path between
any two stations
(note that, for Ethernet in general, use the "5-4-3 rule":
between any two stations, at most 5 segments, 4 repeaters,
3 coax segments. In a pure coax network you are limited
therefore to a 3-segment and therefore a two-repeater rule.)"
Quote snipped from a 2 year old message on this very newsgroup, posted
by Aaron Leonard. The 10base2 ethernet I'm currently on is definately
over this - there are various systems with as many as 6 segments, 5
repeaters, and 3 coax segments between them (10baseT -> hub-> coax ->
repeater-> fiber-optics -> repeater -> coax -> repeater -> fiber ->
repeater -> coax, in the worst case of which there are about 6 nodes on
the 10baseT hubs). The network does work, amazingly enough, but it's
sustainable throughput maxes out at about 700kb/sec, and is often less
than 600kb/sec. Fortunately, our upgrade to a fully switched 10/100-base-T
network started last Friday (Yea! Although we now seem to be entering the
Hell of Too Many Cables).
--- Carl