[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

4208.0. "Does BMS require 32MB ?" by BELFST::ROONEY (Hugh Rooney) Tue Dec 08 1992 05:48

Hi,

I have recently been trying to solve MCC/Motif performance problems on a 
customer site. The problems are related to the user interface e.g. pull down 
menus are slow to draw and the system is slow when moving from one window to 
another. As part of the investigation of this problem I ran the MCC_AUDIT.COM 
procedure which told me that 'a production MCC system should have 32MB of memory'. 
This appears to contradict the SPD which says that 24MB is a valid BMS 
configuration? Can anyone clarify whether 24MB is in fact a configuration which 
we should recommend ?

Has anyone else experienced the slow user interface described above, and is the
addition of memory likely to improve matters. The system is a Vaxstation 3100
M38, it is dedicated to MCC and is not used very heavily (very few alarms
etc.).

Many Thanks for any help/suggestions 

regards

Hugh Rooney 
T.RTitleUserPersonal
Name
DateLines
4208.1What are your Working Set parameters ?ONTIME::BUGSBY::COBBWriting from ALPHA AXPTue Dec 08 1992 07:1130
    I had this problem even with 48 meg of memory. I am running it on a
    3900, which could be argued as a unsupported BMS platform according to
    the SPD.  I bumped up the quotas on the account it runs out of and 
    performance has improved.  My quotas are below:
    
    Maxjobs:         0  Fillm:      1000  Bytlm:       100000
    Maxacctjobs:     0  Shrfillm:      0  Pbytlm:           0
    Maxdetach:       0  BIOlm:      1000  JTquota:       2048
    Prclm:           0  DIOlm:      1000  WSdef:        28000
    Prio:            4  ASTlm:      1000  WSquo:        30000
    Queprio:         0  TQElm:      1000  WSextent:     34000
    CPU:        (none)  Enqlm:      1000  Pgflquo:     150000
    
    The one that made the real difference was the Working set parameters.
    Now with 24 meg, you may not be able to go as high as I have it.  I
    have seen my MCC process grab as much as 30000 of the WSEXTENT.  Before
    I bumped it up, I would see my MCC process constantly maxed out at
    WSEXTENT.  That is when I had slow menu pulldown problems.  I moved it
    up and the problems have gone away.
    
    How many other things do you have going on other than MCC?  How many
    alarms do you have running?  Remote sinks and events turned on?  All
    of these things take more processing power away from the interface.
    
    With 50 alarms and exporting going on and 5 LTM interfaces running, 
    performance is not too bad.
    
    Hope this helps,
    
    Bill
4208.2MCC_AUDIT is reasonableFARMS::LYONSAhh, but fortunately, I have the key to escape reality.Wed Dec 09 1992 20:4214
MCC_AUDIT makes suggestions for REASONABLE performance.

While the SPD does say only 24 Meg is the MINIMUM, please note that it is
the MINIMUM, e.g. the ABSOLUTE MINIMUM, and don't ever configure a system with
that little memory if you expect it to run in a reasonable fashion.
In your base note you said:
	a production MCC system should have 32MB of memory
	  ^^^^^^^^^^
Thats the point.  In the future, I would not be at all supprised if the numbers
were raised significantly in the SPD, like they were for the EMS kit.

If you want to know what size of memory to have on a system,  run the MCC sizer
tool pointed to in the NOTED::MCC-TOOLS conference.  It does a reasonable job
for V1.2 based systems.
4208.3Many ThanksBELFST::ROONEYHugh RooneyThu Dec 10 1992 05:4318
re: .1 

Thanks for the info Bill, I have tried this approach and the performance has 
improved.


re: .2 

I would have thought that a minimum configuration was one that did offer 
reasonable performance. The system performance is now just about acceptable 
after the changes outlined in .1. I would however agree that the 24MB minimum
in the SPD should be reviewed. 

Thanks for the pointer to the sizing tool, it is very useful. 


hugh