Revamping sro database

I was having a look on the released database (Taiwan sro for example), and I was thinking about making new database schema, but I'm confused about how efficient it gonna be. What I was thinking about is:
- Cleaning up the released database from unused tables/procedure.
- Patching the server files or creating a new layer to accept the incoming request from server files, so they are execute only procedures (I found that sometimes they execute procedure and sometimes just direct queries) to make sure that the server files don't conflict with the new schema.
So I would you to share with you how it gonna be from your opinion.

Why I'm going to do that? I found that the current DB is too messy to even track a problem, also I was thinking about creating an open source project to simulate the current working server files, so I was thinking of starting from just simple bright side :D.

I think that rewriting the stored procedures from using cursors to loops might have a relatively big impact, but can't know until it's done and on a server with more than 800 concurrent players to be able to note a difference.

The schema of sro is not so bad, but the items index gets bulky and it favors having unused items data in the database which accumulate to millions.

There's a guy who proposed improvements on stored procedures to delete items instead of flagging them as unused, but you also need to write a query to remove the existing unused items to be able to notice the improvement which was not achieved for storing so many data.

I wonder how many of the big servers rewrote their shard stored procedures.