No subject

a full duplex connection using fiber.  Plenty of bandwidth, and as
long as the load is kept at ~ 30% or less of the ethernet we should
see no problems with jitter in latency.  Gigabit ethernet is point-
to-point, requiring us to use switches at the point where the
antennas converge on the control building so there are no collision 
domains between a given antenna and the control building.  

Within an antenna - again switched ethernet, with each MIB wired
into a switch port, so, again, no collision domains.  Using a
worsksheet produced by a company which specializes in using ethernet
in real-time applications I pluggged in some numbers which read
as follows - network bandwidth, 100Mbps, packet size varied from
8 bytes to 1024 bytes, by powers of 2, message rate varied from 
5000 pkts/sec to 2000 pkts/sec, a deadline for delivery of any
given message of 1ms, and a system lifetime of 20 years.
The result was network loading factors of from 0.32% to 8.19%
with a probability of not missing the 1 ms deadline over the
20 years which varied from 0.99997 to 0.91797.

> 
> It seems the system is architected as one computer to control the whole
> array, instead of a computer per antenna.  This will make your life
> much more difficult, I think.  Why was the distributed approach thrown
> out?

It hasn't been.  One computer to control the whole array is a
misimpression.  I would love to write more on this topic, but
I'm out of time right now.  Several other issues press, so I
will have to defer a reply on this point.

> 
> Please ignore the questions if they're answered in the design
> documents on the Web.
> 
> John
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss