[Pharo-users] Analysis/Visualizations of DBs

Stephan Eggermont stephan at stack.nl
Fri Oct 2 10:24:09 EDT 2015

On 02-10-15 14:34, Torsten Bergmann wrote:
> What I think of is a tool where (after connecting to a possibly unknown database) you can easily
> get/build an overview on the design, quality and data of the database (similar to what Moose, ViDi does
> for code).

In general, this is difficult to do. Our tools assume that all data is 
in ram and you can easily calculate the metrics you want to display.
That means that currently it works well for databases < ~2GB. With a 64 
bit VM we can easily use ~500 GB in a workstation (let's see what the GC 
thinks of that) or 60GB in a cheap PC.

For analyzing legacy systems, the current limits are enough to analyze 
and migrate an ERP system for a 100-people company with 20  years of 
production data. What can be extremely helpful is normalizing data while 
reading it from the database. You'll often find that you can reduce file 
size for the largest tables by a factor 10. To get a quick idea of the 
normalization of the data, try zipping a dump file.


More information about the Pharo-users mailing list