|
Tom,
What protocols are you planning to deploy? LAT, Telnet, Direct?
Combination of two or all three protocols available? Which terminal
servers are you going to use? Are the servers dedicated to Console
functions only?
Bob
|
| We are planning to use the POLYCENTER Console Manager (CM) V1.0 software
running on one or more DECstation 5000/240 workstations that speak
host-initiated TELNET sessions to DECserver 700 ports connected to the
system consoles of VAXcluster VAX 6000 class members, HP-MPE, Prime
(PRIMOS), Convex, and Tandem systems. The DECstation 5000/240s
running the CM software will reside in a Central Management Center
connected by a single T1 line to the customer's network which looks like
the following:
CMC
DS5000/240 DS5000/240
| |
==================================================== Ethernet
|
Cisco
|
| T1 Line (TCP/IP)
|
Cisco
|
DECconcentrator
|
|---------------------------------------------------|
| |
| FDDI backbone |
| |
|---------------------------------------------------|
|
DECconcentrator
|
LAN 10/100 Bridge
|
Cabletron Hub
|
================================================== Ethernet
| |
VAXcluster ---------- DS700
(VAX 6xxx/HSCs) RS232C
Approximate system counts are as follows:
13 VAXclusters (VAX 6xxxs, 2HSCs per Cluster, 1-7 members each)
1 HP-MPE
1 Convex
1 Tandem
25 Primes
----
119 systems total worst case
It is difficult to assess the impact of console message traffic on the
above network, especially during bursty console message traffic periods.
For example, console messages from one VAXcluster were generated every
.06 seconds for a two minute time interval until the problem was
corrected. I am assuming that the console message rate will be the
worst for the VAXes and very little for the other systems.
Assuming no network backups, please respond to the following questions:
Questions:
1) Will the single T1 line between the CMC and FDDI be enough bandwidth
for the aggregate load, especially during bursty console message
traffic periods?
2) Will a single CM engine (DECstation 5000/240, 96MB memory, 4 1GB
disks) be sufficient or should I use two DECstation 5000/240s in the
CMC?
3) If a VAXcluster is remotely located from the FDDI ring over a WAN T1
connection between two Cisco routers, how much T1 WAN bandwidth should I
expect to be consumed worst case?
4) Is anyone doing something similar to the above with VCS? If so, what
network load are you experiencing?
5) What cost-effective contingency plans would you suggest if a network
segment has too little bandwidth?
Please advise...anyone?
This note also appears in the VCS and VAXcluster conferences.
\Thanks in Advance
|
| How quick is a T1 line?
Are all the systems remote from the central command centre? If so, why
not put the DECstation there, and window back to the command centre?
Certainly, if bandwidth is an issue, put the DECstation where the most
systems are.
Martin
|
| Martin,
with the current implimentation which does not have dynamic failover
between systems I would suggest the best place for the CM host is on the central
managed site. I agree there is a network overhead on sending either the console
or the X-windows, but I feel that the main controlling system should be where
the experienced personnel are.
Cheers SImon.......
|