Re: ER/Win vs Designer/2000 - Which is better?

From: Roger Snowden <rsnowden_at_IHateSpam.com>
Date: 1997/04/02
Message-ID: <01bc3f7a$0464b4c0$096fc589_at_RSNOWDEN.labinterlink.com>#1/1


I certainly can't resist replying to this one. I have been using ERWin at several shops for the past two years. A fine product, indeed. It does a good job of physical modelling, schema generation, trigger code generation and a reasonable job of reverse engineering. You have to practice with it, though. I have seen a couple of cases where a schema was generated that was just plain silly. In each case, the designer / user was not particularly familiar with normalization and simply let the tool make stupid decisions. For example, a primary key on a child, the concatenation of two columns, A and B. Then, the tool generated another index on A alone, since it was a foreign key. The second index is pointless, of course. Stuff like that. You have to review the generation scripts carefully and be willing to blow away a schema and start over. It has pretty good support for multiple dbms products, and I have used it to support Oracle, MS SQL Server, Sybase and Access. No sweat on any of them.

Designer/2000, on the other hand, is magnificent. It is a resource hog, but I simply didn't care. It needs an Oracle repository. You must have an instance available with maybe 100 Meg of tablespace. Put the repository in its own tablespace, to keep it clean.

I plopped it on a P166 with NT 3.51 and 64 Meg. I put a local copy of Oracle alongside it and stuck the repository in there. I also built a repository on a DEC Alpha / Unix server when I understood the product better and was ready for ongoing development.

D2k does logical modelling, which ERWin does *not*. ERWin is a physical modelling tool only. D2k also does reverse engineering and - get this - allows you to make incremental changes to a database. It will detect the differences between two schemas and generate a script that will alter one to match the other. That means you can do ongoing maintenance with the tool. With ERWin, if you want to add a column to a table, the tool [Quoted] generates a script to recreate the table. D2k will generate the 'alter' statements. May not sound like much, but if you need to get control of an existing environment, or need to understand the differences between a development and test instance, you have to have D2k. D2k has come a long way.

If you are going to use it with a remote repository, 32 Meg should be just fine. You will need either Win/95 or NT on the workstation.

It also does process modelling and has some code generation tools for C++ and VB, etc. I haven't messed with that part of it yet.

Hope this helps a bit.

Roger Snowden

>
> Entity-Relationship Diagrams
> Logical Modeling
> Physical Modeling
> Maintaining a Data Repository
> Automatically generate Oracle DDL from the Repository
> Reverse Re-Engineer Oracle Tables back into the repository
Received on Wed Apr 02 1997 - 00:00:00 CEST

Original text of this message