Is it possible to build a Data Dictionary

[HerHess]

Is it possible to generate or build or document a data dictionary. Using the table definitions or creating a new table with all the unique fields or creating a new type and steretype.
 

[Bruno.Cossi]

Hi,

what exactly do you mean by data dictionary? Data dictionary in it's most typical meaning can be described/documented/generated using the Class Diagram with the appropriate stereotypes. Or are you referring to something else?

Bruno

Is it possible to generate or build or document a data dictionary. Using the table definitions or creating a new table with all the unique fields or creating a new type and steretype.
  

[HerHess]

Oh, thanks a lot Bruno. The thing is. I have imported DDL Schema from ODBC to have table definitions from a database, now i have my tables, columns, and attributes, what i want is to document the meaning of every field, i can open every table and document every column, but in a relational model, there are redundancy. So i thought the possibility to document only once. Just like in a data dictionary. Thanks a lot for your patience and time.

[Bruno.Cossi]

Hi,

oh, I see what you mean. That is a bit of a problem - in a traditional database modeling world, you would have a logical and a physical model. The physical model, being a mirror of the actual database, indeed does contain a fair amount of duplication, but a logical model does not (at least not when it is designed properly).
And therein lies the problem - EA is not a database modeling tool, and as such does not support the logical database model (not in an automated way, anyway).
So, the short answer would be no, I am afraid.

The long answer - obviously depends on the depth to which you want to document this, but typically a meaning of two fields of a same name, but in two different tables, will be somewhat different. The column definition is typically valid within a context of a table. So you still may want to describe the definitions for each of the columns, even the duplicated ones.

Just out of curiousity, what is the degree of duplication you are seeing in your database?

Thanks,
Bruno

P.S. Do not thank me for my time and patience, time I am happy to share when I have it, and patience I was born without :-)

[HerHess]

Well, there are several types of redundancies.

- We have three subsystems with three model databases, with redundant entities, because the design of independency in case of failure in every machine. So the same info is in three diferent databases for some tables.
- The central repository contains summary tables and statistics tables, with redundant key access.
- The main subsystem has several main entities, por example the main table of tp (telephone) has the unique identification field, this field appears in the uc table (conexion unit), and tp_changes, and tp_deleted, and count_d_tp, and, count_m_tp, and count_ed_tp, and count_tp_all, and others; happen the same with uc table.

We have another question: right now we are using the MS ACCESS DB as repository of EA, but we want to change this repository for INFORMIX or MySQL or other in UNIX. Do you know if this is possible and how.

Thanks and have a good day.

Alberto.