Conference azur::mcc

So you're CPU-bound, not memory-bound?  That seems odd for a 4000 class
system...   

Out of curiosity, how much memory do you have?

Did you install the DECmcc V1.2 mandatory update?  If you haven't then it 
should fix some of the performance problems...

It *does*sound like you have a lot of exporting going on... Did you put all of
the objects you're exporting in a single domain? Or at least try to *minimize*
the number of domains you're exporting from?

BTW:
If you decide to wait for DECmcc on Alpha/VMS , you'll be waiting a while....

/doug

decmcc ssb 1.2
registered entity node4 100
 '"               circuit 70
alarms function with 30 entity dna4,4 polls per hours.
data collector 300 event per hour
sinking decnet event with 300 per hour.
exporter 180 dna4 entity ,4 polls per hour
historian 180 dna4 entity ,4 polls per hour

we are using vax 4000/60  vms 5.5

-----------------------------------------------------
Are you EXPORTing and RECORDing on the same entities? At a glance from the 
above info it looks like you are. You should only need to do ONE of the two.
If this is the case pick one and use it exclusively. Also the historical polls
are every 15 minutes, do you absolutely need this granularity 7x24 for all
entities? We only poll hourly for long term trend analysis and if an individual
entity gets into trouble we may set something up more granular just for that
entity. EXPORT probably consumes lots more of everything compared to RECORD so
if you are really in a bind make do with RECORD. For example, in order to get
NODE4 stats you have to RECORD node4 ckt counters (hourly), ckt char (daily),
and line char. (daily). This would be 26 polls per day per ckt. If you EXPORT
hourly you will poll for ckt counters/status/id/char/ref/etc hourly. This is
probably about 6 times more polling for each entity and statistics will also
be calculated each interval (each entity) costing CPU cycles.

this help??
brad... 

What did the system sizer tool say bout cpu and memory requiremnts?
I would be interested in what it said about your system....

    
    What is generating 300 DECnet events per hour?
    
    What kind of events? We filter most events here (ie. node unreachables,
    adjacency up and down, etc) partly because of overrunning the system.
    
    What is the data collector doing?
    
    Dave

This project concerns the army ,it was signed 3 years ago and we make quite
a lot of mistakes not knowing what DECMCC can do and what hardware DECMCC
need.

ans.1
No I didn't installed the v1.2 MUP,I'll do it Sunday.
I have only one domain for the HISTORIAN.I did try to get the EXPORTER
from the past but it take ages.

ans.2
Yes you are right it's kind of stupid think to do HISTORIAN and EXPORTER
in the same time for the same entities .The reason is that we have allot 
of WAN and the error percent on these lines is quite high (that's why
we poll the counters each 15 min and char each 14 day).The customer want
to be able to analyze the last seven day from the MIR and get report for
one month throw the RDB.We promise to give him this two versions.

Do you believe the 4000/90 will help??

Thanks
Yael Kalai                

ans. 4

; What is generating 300 DECnet events hour?  ;

The customer during the acceptance tests, most of the events are Circuit Up
Circuit Down.

The data collector is used for the DCM events and for painting Circuit
entity for Error ,Utilization ,DECnet events.

Yael