T.R | Title | User | Personal Name | Date | Lines |
---|
708.1 | Line card dependent. | QUIVER::GALLAGHER | | Tue Feb 08 1994 13:27 | 27 |
| >1. Are there some restrictions to assign the six Ethernets on the backplane,
>or can I use any of the 15 IMB's including the half-ones? (IMB11a,11b and
>12a,12b)
The current MAX modules (DECrepeater900TM, DECserver900TM) are only capable
of connecting to the thinwire and matrix segments 1 thru 6. This is due to
a hardware restriction (cost savings) in the line cards. It is not due to
any restriction in the backplane hardware, or in the Hub Manager's matrix
allocation code.
In theory, there can be an Ethernet line card which can connect to any of
the backplane LAN segments.
>2. Are there some restrictions to assign the FDDI Point_to_Point connections
>on the backplane or is it possible to use all the IMB's?
Any of the backplane segments can be used for FDDI. The upcoming FDDI
concentrator line card can connect to any of the backplane matrix segments.
FYI: The matrix allocation algorithm allocates Ethernet segments from
segment 1 on up, and FDDI segments from 14 down. (If two FDDI cards to
be connected are close enough to one another to use the split channels,
these are used.)
Good luck on the tool. Feel free to post a pointer to it so we can
try it out.
-Shawn
|
708.2 | matrix segments not clear | GENIE::MICHEROLI | | Tue Feb 08 1994 14:58 | 9 |
| Thanks a lot
"matrix segments 1 thru 6" means 1 IMB on the upper connector and 5 IMB's
on the lower connector. Is that right?
PS
By the way, is there a technical manual avaiable for DEChub900 and its
modules?
That would help a lot (for a free configtool demo!)
|
708.3 | Too many synonyms. | QUIVER::GALLAGHER | | Tue Feb 08 1994 15:43 | 20 |
| > "matrix segments 1 thru 6" means 1 IMB on the upper connector and 5 IMB's
> on the lower connector. Is that right?
Yep. I should have phrased it as, the flexible channel on the upper connector
and the first 5 lower connector channels.
> By the way, is there a technical manual avaiable for DEChub900 and its
> modules?
The only documentation I've seen on the backplane matrix is the "System
Spec", by Rich Graham. It's available from:
quiver::PROJ$722:[ONEHUB.SYSTEM]ONEHUB_SYSTEM_SPEC_<latest version>.PS
This area requires access rights. Contact Rich at levers::graham if you
have trouble accessing it.
I don't know of any document which describes each product's backplane
connection capabilities.
-Shawn
|
708.4 | | LEVERS::SWEET | | Wed Feb 09 1994 13:21 | 6 |
| I would be interested in the work you are doing. I would like to
understand its relationship to HUBwatch.
Please send mail to LEVERS::SWEET with details of your plans.
bruce
|
708.5 | A note of caution... | QUIVER::SLAWRENCE | | Mon Feb 14 1994 14:06 | 32 |
|
This needs some correction.
There are 120-some 'signals' on the 900 backplane, which modules group
to form 'channels'. Each type of module can have its own definition of
what constitutes a 'channel' (as it happens, all current modules use
the same groupings, but future modules will use others that overlap
with the current ones).
The current Ethernet full-height modules use the same channels; one
signal set on the upper connector and 5 on the lower connector (the new
half-height repeater will not be able to use the 5 on the lower
connector). This is the source of the restriction that there can be
only 6 backplane ethernets. There are enough signals to make the
theoretical maximum 15 + the thinwire, but this was not done for cost
reasons.
Note that when you use HUBwatch (or any other SNMP manager that knows
the correct MIBs) to configure the backplane connections you _cannot_
select which channels are assigned. All you can do is tell the MAM
what network segment types you want created and what backplane ports
you want connected to each.
If the configuration you request is possible given the signals
accessible by each module, the MAM will compute an allocation that
creates the topology that you want, but even then you can't tell which
physical signals are being used for which segment.
The allocation order described in one of the earlier replies is an
artifact of the capabilities of the current repeaters and _will_ change
for configurations that will be possible with newer modules.
|
708.6 | | CSC32::MACGREGOR | Colorado: the TRUE mid-west | Wed Jul 26 1995 14:36 | 9 |
|
I have a followup question concerning IMBs. Is there any hope that we
will be able to put more than 6 ethernets and solve the ADD FAILED
problem any time in the future?
Marc
|
708.7 | new location ? | LEMAN::CHEVAUX | Patrick Chevaux @GEO, DTN 821-4150 | Tue Jul 30 1996 05:02 | 11 |
| .3�> By the way, is there a technical manual avaiable for DEChub900 and its
.3�> modules?
.3�
.3�The only documentation I've seen on the backplane matrix is the "System
.3�Spec", by Rich Graham. It's available from:
.3�
.3� quiver::PROJ$722:[ONEHUB.SYSTEM]ONEHUB_SYSTEM_SPEC_<latest version>.PS
Node QUIVER is unknown. Any info on where I could locate the specs ?
Thanks and reagrds, Patrick
|
708.8 | Use NETCAD:: instead of quiver:: | NETCAD::BATTERSBY | Don't use time/words carelessly | Tue Jul 30 1996 09:48 | 5 |
| Quiver used to be part of the NETCAD cluster, so if you substitute
"quiver" with "netcad", you should be able to get at the document
you're looking for.
Bob
|
708.9 | NETCAD | LEMAN::CHEVAUX | Patrick Chevaux @GEO, DTN 821-4150 | Tue Jul 30 1996 13:12 | 3 |
| Bob, thanks for your prompt response. NETCAD is alive but the account
seems to be protected as I can't dir into it. Do you know any other
location for those specs ?
|
708.10 | Send mail to Rich about this & ask him... | NETCAD::BATTERSBY | Don't use time/words carelessly | Tue Jul 30 1996 13:31 | 7 |
| No Patrick, I do not know. These are engineering specs, and as such
need to be protected from general access. There is I believe
an access control list for access to this area.
I'd suggest you send mail to Rich Graham and inquire about access
and state reason for your need.
Bob
|