[comp.sys.proteon] New hardware and software

jch@DEVVAX.TN.CORNELL.EDU (Jeffrey C Honig) (05/23/88)

Can anyone relate any experiences, good or bad, with Proteon's new
Ethernet and CPU cards and version 8.0 of the software?

Does anyone actually have any of these new goodies yet?

Thanks

Jeff

lekash@ORVILLE.NAS.NASA.GOV (John Lekashman) (05/23/88)

We've had 8.0 running on a pronet-80 to ethernet gw for about
seven weeks.
It hasn't crashed or lost it at all. 
Works fine.  I did have to restart it once 5 weeks ago
when a static ram parameter changed.
Proteon did a fine job on the installation, redid  our whole
static ram config for us on it.  Whether your installer will
come shoot me for telling you, I don't know.

End to end throughput is at least double what it was in 7.4.
Some numbers for using 8.0:
sun 3/260 on ether to sun on pronet-80 290 kBytes/second
sun 3/260 on pronet-80 to sun on ethernet 210 kbytes/second

With streams running both ways, the GW handles
1400 packets/second.  This is about half 'large' packets (1K for ftp data)
and half small (64 byte ack packets.)  It is not certain whether
this number could be larger when more hosts get involved.
Maybe I'll know later.

I haven't yet accurately measured aggregate throughput,
but we can get at least an effective 2.3 mbits/second through it
into host machine memory, using TCP.  This is the effective
data rate, including acks.  From the way timing results look,
I suspect aggregate from multiple hosts is larger.

					john

RAF@NIHCU.BITNET ("Roger Fajman") (05/23/88)

We've been running 8.0 on our gateway (without new CPU or Ethernet
cards) for a couple of weeks now with no problems.  The upgrade
(which includes various new ROMs) wouldn't work when I installed it,
but did when Ken Crocker from Proteon came here to do it.  We got
8.0 rather early because Proteon could never get 7.4 to work on our
gateway, which has a lot of interfaces.  Not having 7.4 kept us from
connecting to SURANET because of the subnetting restrictions in 7.3.
Since 8.0 was installed, our T1 line to the University of Maryland
for SURANET has been working fine, but is not yet heavily loaded.

Our gateway has 11 interfaces, 5 of which are actually in use at the
present time.  We do plan to upgrade to the new CPU and Ethernet
cards.

kwe@bu-cs.BU.EDU (kwe@bu-it.bu.edu (Kent W. England)) (05/23/88)

In article <8805230110.AA00698@devvax.TN.CORNELL.EDU> jch@DEVVAX.TN.CORNELL.EDU (Jeffrey C Honig) writes:
>Can anyone relate any experiences, good or bad, with Proteon's new
>Ethernet and CPU cards and version 8.0 of the software?
>
>Does anyone actually have any of these new goodies yet?

	Boston University beta tested the new CPUs and new Enet boards
with good results.  We didn't have any hardware problems to speak of
and only one software problem.  To save memory, Proteon changed the
default number of DECnet nodes and  this caused us to lose some DECNet
nodes until we figured out the source of the problem.

	I don't have any hard throughput numbers.  We are running IP
and DECnet.  Rel 7.4 worked well for us, so Rel 8.0 was transparent.

	There are a few added features in 8.0.  One nice one is
console port security.  There are also configureable ARP cache
time-out parameters now.

	I hope to have more hard data soon, but we have been
overworked and struggling to fix the Pronet FOI receiver aging problem
at the same time as the beta.

	Kent England, Boston University