Hello everybody,
I am currently using QlikView Personal Edition to do some analysis on the inventory of my wholesale business. The data comes directly from my ERP system, a huge Oracle database with a few thousand tables and overall several million rows. As I need a solution that can be used by a whole team, I found Metabase, but I am quite lost on how to start.
I connected the Oracle database and after some time, my tables show up. Do I have to edit all the tables manually now, set most of them to invisible to get a usable tool? Or is there a way to specify the tables to import manually? For the old approach I do have SQL statements that generate for example a table of items, their names and important values for analysis (coming from around 20 tables). Can I use them as starting point?
Or should I add these statements as views to the original database for analysis?
Any help is appreciated.
Regards,
Christoph
Hi @RedTiger
I have seen much bigger instances, but the table count would be overwhelming for most users, so hide everything that isn't relevant.
You would have to manually hide everything that you don't want - or use a database user, who only have access to those specific tables (I'm not sure if Oracle can handle that).
But you can hide all tables with a single click (in a database/schema) and then unhide the tables that you want to expose:
https://www.metabase.com/docs/latest/administration-guide/03-metadata-editing.html#table-visibility
There's a request for limiting schemas/tables to sync:
https://github.com/metabase/metabase/issues/5500 - upvote by clicking 
on the first post
You might be interested in some of the linked issues in this comment too:
https://github.com/metabase/metabase/issues/2146#issuecomment-631995455
I would generally recommend using Views if possible, since it's much easier for the users to start creating questions from that compared to SQL, which is still possible, but they would have to create questions based on that SQL question (via Saved Questions).
https://www.metabase.com/docs/latest/users-guide/custom-questions.html#picking-your-starting-data
It's kinda "where do you want to put the burden" - on you or the users?
Thanks alot for the ideas.
I did the following now: I created a user/schema in Oracle that holds views with the data that is relevant for metabase. I organized the views in way to simplify the data structure for the users (so there is one table for my customers, one for sales items, ...). The user needs the "select any table" and "select any view" permission to do this smoothly.
Then I created a user/schema metabase which has only access to those views from the first user. The metabase user is now used to connect and so the data is clean and simple and my users can start working with it.
Works like a charm.