Conference azur::mcc

1812.0. "Modelling managed object rules?" by RIVAGE::SILVA (Carl Silva - Telecom Eng - DTN 828-5339) Thu Nov 14 1991 07:33

	It seems like there are always problems with the following cases of
modelling managed objects:

	1. The use of too many large ASCII strings that are user modifiable 
usually means that further investigation should be made into the attribute to 
determine its applicability or its usage in the managed object.

	2. The use of complicated attributes may indicate the need for a new 
object class that could simplify the model and provide greater flexibility.  
For instance, all the attributes describing a person could be contained in a 
Record data type or an attribute could be used that contains the identifier to 
point to a  new object class called User Profile.  This new object class could 
contain all of the information contained in the Record data type without
the unnecessary duplication of information throughout the managed objects.

	Is there a general rule of thumb handling these cases?  I.e., when
dealing with large ASCII strings, only include them if they are absolutely
necessary or break them down into individual attributes?  Or, in general when a
record data type contains more than 6 or 10 attribute fields it may be better
to reate a child object class or new related object class?

	Any help, feedback, comments on this would be greatly appreciated!

	Carl

	Cross posted in the Enterprise Management Architecture notes file, 
note 105.0.

Answered in the EMA notes conf - most questions on entity design should be
posted there. This is really the wrong conf for them...