| Title: | DEChub/HUBwatch/PROBEwatch CONFERENCE |
| Notice: | Firmware -2, Doc -3, Power -4, HW kits -5, firm load -6&7 |
| Moderator: | NETCAD::COLELLA DT |
| Created: | Wed Nov 13 1991 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 4455 |
| Total number of notes: | 16761 |
Hi,
we have customer using DECHUB 900 and 8* DECmau 900TL. They have
reported some performance problem. They have 16Mbps TokenRing with
original IBM PCs and Novell. They have some tool to generate load in the
network. We made following experiment:
1/ Three stations connected to the ring generated load about 810 kBps =>
about 270 kBps.
2/ Seven stations connected to the ring generated load about 880 kBps =>
about 125 kBps.
3/ In both cases stations were connected through one DECmau.
4/ When we connected stations through several different DECmau results
are the same.
We were told that with the previous installation they have much better
performance (they claim that application's response time was 3-5 times
better). They told they were using the same PCs and Token Ring cards
and different hub and mau (probably IBM).
Our problem is that we don't have any expierence with TokenRing and
even datascope for it (or TR card able to work with IRIS).
Thank you in advance for any help
Jarry Pilar
Digital Services Czech Republic
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 748.1 | Not obvious what the cause is | KALI::SOHA | Mon Feb 21 1994 14:01 | 24 | |
Jarry,
Thanks for the note.
I've sent a short response out via Mike Bouchard. It goes soemthing
like this -
The reason for the performance hit isn't obvious. The DECmau's and
the DEChub 900 use passive technology for intercconection stations on
token ring - relays. Relays will not impact performance.
Two possibilites are:
1. One or more of the stations is at the limit in terms of
signal strength or possibly oscillator stability thus
causing the station to insert, de-insert, insert, etc.
If this was happeing you should be seeing errors on
the suspected station and soft errors on the ring.
2. Something has changed between the two tests. If we
could have the customer re-run using the original
equipment that wopuld be great - or get the specifics
to us and we can look into re-creating it here in the lab.
Please keep me in the loop.
Thanks Mike S.
| |||||
| 748.2 | Nothing changed between tests | DECPRG::PILAR | Tue Feb 22 1994 03:18 | 11 | |
Hi,
I do not think something has changed between tests. They were performed
in several minutes. I'll try to investigate more preciously former
configuration to allow you to simulate it in lab. I was asking
customer about error counters etc. but they told me there is nothing.
It seems to me there is some "magic" border (900kBps).
Regards
Jarry
| |||||
| 748.3 | Good news from Prague | DECPRG::PILAR | Mon Feb 28 1994 10:19 | 20 | |
Hi all,
we finally resolved this issue. We made several further experiments to
isolate problem. We have PCs connected directly to MAU without DECsite
cables, we were changing stations participating in test. Nothing has
helped - behaviour was always the same. There was only one thing we
cannot overcome - server. It was not possible to run test without
server. So server became suspicious. Because we do not have any test
equipment for TokenRing (which is quite rare here in Czech republic) we
propose to customer to test another vendor MAU. The have borrowed one
from IBM dealer in Prague and results were fortunately the same. They
have called us today and it seems that problem is in Novell server
(probably TokenRing card).
It is now quite good for us because we will probably be able to sell
additional DECsite for $250k.
Regards
Jarry
| |||||