Can EA Really Do Data Modelling

[chaluwa]

We are taking a good look at the range of UML modeling tools currently available, EA seems to stand out but I am left with one question.
Does EA also extensively handle data modeling (design, reverse engineering, export, import etc) or should we consider other specific tools like Toad Data Modeler or those from IBM (Rational Data Architect / Rational Data Modeler) ??. I really need honest responses. Cheers.

[«Midnight»]

Unfortunately, your question is one of those where the answer is "it depends..."

Yes, EA can do all that you describe. But...

No, EA does not do all these things at the level of detail and functionality that higher-end tools like Toad can provide. One area you might find EA weak on is full round trip engineering. EA can provide only limited support for changing the DDL of a preexisting schema. Something like Toad takes this completely in stride. Toad (et al) also has far stronger support for views and such. Finally, special purpose tools often provide much better help for stored procedure creation and editing.

You can search my older posts on the Data Modeling Profile for some background, but that might be more academic than practical for your choice.

My advice is to download the EA documentation (which has the same content as the help file) and look through the data modeling section. EA can do a bit more than is mentioned there. The documentation is still catching up a bit; in the past 24 hours it became apparent that Oracle triggers were not fully documented, as noted in another forum thread. Still, that's your best bet.

Even if you continue to use Toad (or whatever) you might still find the EA models useful. EA could import the schema, then provide a means to relate software design to your data structures, propose high-level changes, and produce unified documentation.

David

PS: Keep watching this thread. There are bound to be other observations (we're a vocal bunch here). Some of the users who work with Oracle a lot will have some more focused comments.

[Michael Leisch]

By "limited support for changing the DDL" do you mean the DDL that is generated will drop the existing objects before it creates them?

Is there any way to save (or even print) the results of the DDL Compare? Even that would be helpful.  

[«Midnight»]

Hi again Michael,

Yes, the DDL will (optionally, depending on the settings you choose) generate DROP statements before creating new objects. What I was referring to was a situation where you modify (say) a table definition, and want EA to create the necessary ALTER statements. With very few exceptions EA does not (yet) have this capability. The 'yet' modifier refers to the stated intentions (from time to time) to extend EA in this direction. AFAIK there have not been firm time frames or specifications for this as yet.

I don't think EA can do a DDL compare. [It's possible I don't quite understand what you are referring to though.]

David

[Keithmat]

I don't think this situation has changed but thought I would ask.

Has there been any improvements to Class ---> Database change class ---> Database engineering?

I can get the forward engineering working fine but if I make any changes to the data model then change a class and forward engineer again I can't get it to stop overwriting the table?

Anyone know how to do it?

[qwerty]

It's the same state as ever. There's some MDG called - uhm - DB<something> which could be of help.

q.

[jfzouain]

Yes, is called DB Mode and is a very good add on for EA Data Modeling.

[Glassboy]

I really need honest responses. Cheers.

It actually depends on who you have doing it.  People are extremely bigoted about tools, and the data modelling space seems to be an area where people are extremely bigoted.

Personally I've found all EA's capabilities in the data space useful.

[icon02#009]

I have been using EA to produce a sort of "hybrid" Data Model / Data Flow that shows also flows between, for example, Applications and API, and both of those and the Data Warehouse. The Applications are predominantly Oracle APEX but also .NET. I have been able to reverse engineer from the Oracle DB schema (about 40) and the .NET Databases (about 20), once I had got over the pain of producing 32-bit ODBC drivers in a 64-bit environment (Windows 10). I don't think Toad would be able to reverse engineer the latter ?
My client is very happy with the Model on a Page (OK - its a very large page ! Bigger than A0) that shows the complexity of the Application Landscape and now wants me to continue adding detail.
There are some limitations in the oracle reverse engineer - it does not bring the columns for a View (although it does bring the SQL) and does not really cope with Materialised Views (although the Table structure does come across the code does not)
I have used EA a lot in the past in the Data Migration space as a way of "getting to know" a Legacy Database and establishing the Daat Mapping to the Target system ( especially if that was also reverse engineered in to the Model)
I , and the client , have been impressed with the Model queries that I have been able to develop as well the rtf reporting.

Everyone will have opinion of what tool is best ? My vote for EA is based on the richness of the possible models that can be produced, as well as the relatively low licence costs , when compared to say ARIS or BiZZDesign.

I am always having to talk further about the work I have done , however "impure" it may be in true Data Modelling terms !!  , just PM me