Conference azur::mcc

4521.0. "Configuration Opinions Needed" by TRCOA::TIPPER ($DEF/INSTRUCTOR SANDY "Kenneth A.") Wed Feb 10 1993 09:41

    I have to teach Network Management using DECmcc (V1.2, VMS platform)
    in a few weeks, and I have insufficient experience and woefully 
    inadequate hardware.
    
    Can anyone tell me whether the following compromise will absolutely not 
    work?  I do not expect optimal (or even "acceptable") performance.
    
    I hope to have four workstations used only as X-windows display servers,
    with four larger VAXen in the background actually running four
    independent instances of DECmcc (all with local namespaces for
    simplicity's sake).  Connection is via ethernet and DECnet Phase IV
    (hope to have TCP/IP available for demo only).
    
      Display servers:
    	1 colour VAXstation 3100 with 8 meg of memory, small disk, acting
    		as a MI-Cluster "diskless" client (local swap/page files)
        2 monochrome DECstation 3100s with 16 meg of memory, adequate disk,
    		running ULTRIX 4.3, DECnet/ULTRIX (phase IV)
        1 colour Flamingo (DEC 3000 -500), 96 meg of memory, 
    		running OpenVMS AXP T1.0-FT4
    
      Compute servers: (plenty of disk space available)
    	1 VAX 8530, 32 meg of memory, VMS 5.5, cluster 1.
    	1 VAX 3600, 32 meg of memory, VMS 5.5, stand alone
    	1 VAX 6000-230, 64 meg of memory, VMS 5.5, cluster 2
    	1 VAX 8350, 20 meg of memory, VMS 5.5, cluster 2
    
    I know the 8350 is too small, but it's what I have.  
    
    Qusetions:
    1. Am I out of my mind?
    2. Will the Flamingo successfully work as a simple display server in
	 front of customers (no release forms or anything)?
    3. Is there going to be such enormous network load between the
    	clustering and the X-windows stuff that the users will be totally
    	frustrated (e.g. repeated failures, large interactive time delays,
    	etc.)?  Nothing else is going on on the network, but all four stations
    	will be doing roughly the same things at roughly the same time.
    4. Which cluster would be better for the VAXstation to be a member of
    	(I am leaning toward cluster 1, with the 8530).
    5. Would I be better trying to run two independent mutually invisible
    	instances of DECmcc on the 6000-230 in cluster 2 and leave the 8350
    	out of it? (I think not: consider the fact that the course includes 
    	teaching them [customers] how to install DECmcc on VMS, and doesn't 
    	anticipate weird workarounds for situations they will not see at home.)
    6. Ignoring the Flamingo and the 20-meg system, is this configuration
    	a) supported and acceptable
    	b) supported but poor performance and awkward
    	c) unsupported but probably workable
        d) unworkable 
    	e) rediculous
    
    
    Comments, warnings, hints, dire predicitions all appreciated, as soon
    as possible.  I am trying to assemble and integrate this mess
    immediately.
    
    Sandy Tipper,
    Digital Learning Services,
    Toronto, Canada

    handled off-line.