[CLUE-Tech] Data Architect

[Jeffery Cann]

Since the topic of database design tools comes up on this list, I thought 
some folks would appreciate this post.

At the March CLUE meeting, we gave away the 2.0 version of Data Architect, 
written by Code By Design and distributed by The Kompany.  Until we gave it 
away, I had not used (or even heard of) this tool.  So, I tried out the demo 
version (http://www.thekompany.com/products/dataarchitect/).  The demo is 
full-featured, but only runs for a short period of time.

Note that only ODBC connections are supported, so you need unixODBC, which 
ships with most Linux distributions.  (http://www.unixodbc.org).  I tested 
the ODBC connections to MySQL and Postgresql.  The third type of model is 
SQL-92, which I did not try.  I did not try any other connections, because I 
don't have other databases running at my house.  Also, the model types allow 
you to use native datatypes for PG, or MySQL and standard SQL datatypes for 
SQL-92.

I am impressed with the stability and features of this product.  In
particular, you can specify all of the attributes for your model and Data 
Architect will generate the DDL SQL statements.  In addition, it supports 
reverse engineering.  Its data is stored in XML format, so it appears to have 
portability (to another CASE tool) in mind.  The user interface is intuitive 
and although the help is quite scant, I was able to figure out things only 
from the UI.  For example, within an hour, I built a 9 table database in 
Postgresql, with primary keys, sequences, indexes, and foreign keys.  Much 
longer by hand (492 lines of DDL code from a pg_dump --schema-only of this 9 
table database).

Although this is a commercial application, I think it is well worth the $39 
(download version).  It is also not open source, so you cannot get the source 
code (darn).  Another downside for some users is that Data Architect is a KDE 
application.  KDE is handy because it runs on Linux, Mac, and Windoze, so it 
could be a native, cross-platform tool for some companies.  Finally, it's 
missing some features which I would consider 'nice to have'.  For example, I 
can save an ERD as a Postscript file.  It would be nice if I could export a 
drawing to PNG (as a documentation reference).  Also, the reference names on 
the association arrows are not movable, as they are in most technical drawing 
/ case tools.  Overall, I give it a B+.

Anyone else worked with it?  Offer opinions?  Suggestions for other database 
modeling tools that run on Linux?

Jeff