[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

Title:DECmcc user notes file. Does not replace IPMT.
Notice:Use IPMT for problems. Newsletter location in note 6187
Moderator:TAEC::BEROUD
Created:Mon Aug 21 1989
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:6497
Total number of notes:27359

2393.0. "Non-stop DECmcc?" by NEWPRT::SCHOOLER_CH (Cheap? thrill...Jump a horse) Fri Feb 21 1992 13:07

My DECmcc non-stop requirements are very strigent.  My client would like to see
MCC running on a two node VAXcluster. The display would be on a workstation. 
If the cluster processor fails, MCC would failover with no noticable effect
on the display except for maybe some slighly longer response times. 

The most obvious way to acheive non-stop is the FTvax.  This is my first
choice; we are looking into ramifications.  As for the cluster platform,  I
would think we would need to add failover capabilty to the MCC code and write
a custom distributed PM that could maintain context during failover.  Either
of these is no easy task, but if we are to manage critical networks, the
network manager cannot a point of potential failure.  Any comment? 
T.RTitleUserPersonal
Name
DateLines
2393.1KYOA::KOCHIt never hurts to ask...Sat Feb 22 1992 10:398
    I can't offer anything but the obvious. Configure the appropriate
    machines, find the difference in cost. If the development cost exceeds the
    difference, you have your choice. 
    
    Please be cognizant of the life cycle cost of these changes. In a VAXft
    solution, there are NO code maintenance costs. Given that an VAXft
    solution is usually only 20% higher than a comparably configured
    dual-host solution, I would go for the VAXft solution.
2393.2Go for FT!!!WOTVAX::HATTOSI think, Therefore I'm paid lessMon Feb 24 1992 08:268
    I have T1.2.4 running on a 410FT and have no problems at all, (except
    those already known)
    
    The FT solution would seem to be the best for this situation. In a
    cluster config. you will go through a cluster transition if you lose a
    node, which will cause your MCC display to hang for a while.
    
    Stuart