|
>>1. I am not too much concerned with keeping control of the system console in case
>>the system goes down, but more with the monitoring of events and triggering of
>>alarms and actions.
>>
>>So, I don't think using a dedicated phone line would be the most cost effective
>>choice. I would rather use an X.25 link for that purpose (of course, it would
>>only work as long as the system and the comms are up and running, but that
>>would be ok).
>>
>>How could this be implemented? I read in some other notes about a "pseudo-termi-
>>nal" feature and about an API. Could these be used for the purpose I have? Does
>>anyone have some similar experience?
I take it that these systems are at remote sites which you have no fixed network
link to. As you have mentioned X.25, can you not run TCP/IP over the X.25 link
and then use TELNET to connect to terminal servers managing your systems.
>>2. The total number of nodes I want to manage is greater than 200. How can I
>>overcome this limitation?
Simply install an unlimited licence and manage your systems, of course if you
log a QAR/CLD/SPR when you have more than 200 systems, we will simply reject it
as you will be running an unsupported configuration.
This limit MAY be upped in a future release, but we would have to perform
extensive testing before that happened which we currently dont have the time for.
>>3. How to size the right system (performancewise) for this number of nodes?
As its unsupported and we have never tried it, I would suggest the biggest system
you can get your hands on. (See the sizing guidelines in one of the earlier notes)
>>4. The total price for the licenses for this number of nodes is a huge figure. Is
>>there another pricing approach that could be used?
You do want Digital to make a profit don't you? This is the only approach, we did
at least give you a price break if you but an unlimited license, compare that with
200 individual licences!
Cheers,
Phil
|
| >>So, I don't think using a dedicated phone line would be the most cost effective
>>choice. I would rather use an X.25 link for that purpose (of course, it would
>>only work as long as the system and the comms are up and running, but that
>>would be ok).
>>
>>How could this be implemented? I read in some other notes about a "pseudo-termi-
>>nal" feature and about an API. Could these be used for the purpose I have? Does
>>anyone have some similar experience?
<<I take it that these systems are at remote sites which you have no fixed network
<<link to. As you have mentioned X.25, can you not run TCP/IP over the X.25 link
<<and then use TELNET to connect to terminal servers managing your systems.
(I just realized my explanation wasn't totally clear...)
The 350 nodes will be in 350 different locations. A permanent X.25 link will
exist, over which we plan ro run TCP/IP. However, there are two major obstacles
to using a terminal server at each site "redirecting" the console port: first,
we plan to use 2000/300's with graphical consoles, and I assume in that case I
cannot have an external console port; second, because that would be the only use
for the terminal server (all desktop devices will be PCs with Ethernet card).
I thought we might be able to have some kind of deamon process trying to capture
the console messages, sending them over the X.25 line (TCP/IP?) to the central
system, and there feed them through some kind of API. Does this make sense?
>>2. The total number of nodes I want to manage is greater than 200. How can I
>>overcome this limitation?
<<Simply install an unlimited licence and manage your systems, of course if you
<<log a QAR/CLD/SPR when you have more than 200 systems, we will simply reject it
<<as you will be running an unsupported configuration.
<<This limit MAY be upped in a future release, but we would have to perform
<<extensive testing before that happened which we currently dont have the time for.
>>3. How to size the right system (performancewise) for this number of nodes?
<<As its unsupported and we have never tried it, I would suggest the biggest system
<<you can get your hands on. (See the sizing guidelines in one of the earlier notes)
The sizing guidelines refer only to memory requirements. What level of CPU
performance would be required from an Alpha/OSF system? An 3000/800S could do it?
Smaller? Larger?
>>4. The total price for the licenses for this number of nodes is a huge figure. Is
>>there another pricing approach that could be used?
<<You do want Digital to make a profit don't you? This is the only approach, we did
<<at least give you a price break if you but an unlimited license, compare that with
<<200 individual licences!
I was not aware of the unlimited licence from reading the SPD. Is it the
QL-MU1A9-AA part #?
Thanks,
Paulo
|
|
Ahhh
OK, you are out of luck with the 2000/300 as there is no way to redirect
the console to one of the "COM" ports on the back (unless you unplug your
mouse and keyboard).
You could create some kind of client/server application which the client
ran on the 2000/300 and the server on the console manager system, the
server could run in a pseudo-terminal managed by Console Manager.
Incidentally, this client/server could be POLYCENTER System Watchdog.
Now, as for CPU size, we can only tell you what we have used, I have no
idea how poerful a 3000/800s is however, if it is more powerful than the
systems we have tested with then you should be OK.
As for licencing, the QL-MU1A9-AA is for ULTRIX and yes, it will allow you
to connect up to the 200 system limit.
The other ULTRIX licence is QL-MU1AC-3B which will allow you to connect 1
system for each licence you purchase.
Hope this helps,
Cheers,
Phil
|