Joining across 3000+ tables - mysql

Question

Ok, strange one here
I have a database for customer data. My customers are businesses with their own customers.
I have 3000 tables (one for each business) with several thousand email addresses in each. Each table is identical, save the name.
I need to find a way to find where emails cross over between businesses (ie appear in multiple tables) and the name of the table that they sit in.
I have tried collating all entries and table names into one table and using a "group by", but the volume of data is too high to run this without our server keeling over...
Does anyone have a suggestion on how to accomplish this without running 3000 sets of joins?
Also, I cannot change the data structure AT ALL.
Thanks
EDIT: In response to those "helpful" restructure comments, not my database, not my system, I only started a couple of months ago to analyse the data

Answer 1

Multiple tables of identical structure almost never makes sense, all it would take is a business field to fix this structure. If at all possible you should fix the structure. If it has been foisted upon you and you cannot change it, you should still be able to work with it.
Select the distinct emails and the table name from each table either UNION ALL or pull them into a new table, then use GROUP BY and HAVING to find emails with multiple tables.
SELECT email
FROM Combined_Table
GROUP BY email
HAVING COUNT(sourc_table) > 1

Answer 2

So, you say you can't change the data structure, but you might be able to provide a compatible upgrade.
Provide a new mega table:
CREATE TABLE business_email (
id_business INT(10) NOT NULL,
email VARCHAR(255) NOT NULL UNIQUE,
PRIMARY KEY id_business, email
) ENGINE = MYISAM;
Myisam engine so you don't have to worry about transactions.
Add a trigger to every single business table to duplicate the email into the new one:
DELIMITER \\
CREATE TRIGGER TRG_COPY_EMAIL_BUSINESS1 AFTER INSERT OR UPDATE ON business1 FOR EACH ROW
BEGIN
INSERT INTO `business_email` (`id_business`, `email`) VALUES (NEW.`id_business`, NEW.`email`) ON DUPLICATE KEY UPDATE `id_business`=NEW.`id_business`;
END;
\\
DELIMITER ;
Your problem is to add it dynamically whenever a new table is created. It shouldn't be a problem since apparently there's already dynamic DDL in your application code.
Copy all existing data to the new table:
INSERT INTO `business_email` (`id_business`, `email`)
SELECT email FROM business1
UNION
SELECT email FROM business2
...
;
COMMIT;
proceed with your query on the new business_email table, that should be greatly simplified:
SELECT `id_business` FROM `business_email`
WHERE
GROUP BY `email`
HAVING COUNT(`email`) > 2;
This query should be easy to cope with. If not, please detail the issue as I don't think properly indexed tables should be a problem even for millions of rows (Which I don't believe is the case since we talk about emails)
The advantage of this solution is that you stay up to date all the time, while you don't change the way your application works. You just add another layer to provide additional business value.