Cascading WHERE clause inside a VIEW with a UNION - mysql

Question

This isn't solved, but I found out why: MySQL View containing UNION does not optimize well...In other words SLOW!
Original post:
I'm working with a database for a game. There are two identical tables equipment and safety_dep_box. To check if a player has a piece of equipment I'd like to check both tables.
Instead of doing two queries, I want to take advantage of the UNION functionality in MySQL. I've recently learned that I can create a VIEW. Here's my view:
CREATE VIEW vAllEquip AS SELECT * FROM equipment UNION SELECT * FROM safety_dep_box;
The view created just fine. However when I run
SELECT * FROM vAllEquip WHERE owner=<id>
The query takes forever, while independent select queries are quick. I think I know why, but I don't know how to fix it.
Thanks!
P.S. with Additional Information:
The two tables are identical in structure, but split because they are multi-100-million row tables.
The structure includes primary key on int id, multiple index on int owner.
What I don't understand is the speed difference between the following:
SELECT COUNT(*) FROM (SELECT * FROM equipment WHERE owner=1 UNION ALL SELECT * FROM safety_dep_box WHERE owner=1) AS uES;
0.42 sec
SELECT COUNT(*) FROM (SELECT * FROM equipment WHERE owner=1 UNION SELECT * FROM safety_dep_box WHERE owner=1) AS uES;
0.37 sec
SELECT COUNT(*) FROM vAllEquip WHERE owner=1;
aborted after 60 seconds
Version: 5.1.51
mysql> explain SELECT * FROM equipment UNION SELECT * FROM safety_dep_box;
+----+--------------+----------------+------+---------------+------+---------+------+---------+-------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------+----------------+------+---------------+------+---------+------+---------+-------+
| 1 | PRIMARY | equipment | ALL | NULL | NULL | NULL | NULL | 1499148 | |
| 2 | UNION | safety_dep_box | ALL | NULL | NULL | NULL | NULL | 867321 | |
| NULL | UNION RESULT | <union1,2> | ALL | NULL | NULL | NULL | NULL | NULL | |
+----+--------------+----------------+------+---------------+------+---------+------+---------+-------+
with a WHERE clause
mysql> explain SELECT * FROM equipment WHERE owner=1 UNION ALL SELECT * FROM safety_dep_box WHERE owner=1
-> ;
+----+--------------+----------------+------+-----------------------+-------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------+----------------+------+-----------------------+-------+---------+-------+------+-------+
| 1 | PRIMARY | equipment | ref | owner,owner_2,owner_3 | owner | 4 | const | 1 | |
| 2 | UNION | safety_dep_box | ref | owner,owner_3 | owner | 4 | const | 1 | |
| NULL | UNION RESULT | <union1,2> | ALL | NULL | NULL | NULL | NULL | NULL | |
+----+--------------+----------------+------+-----------------------+-------+---------+-------+------+-------+

Answer 1

First off, you should probably be using UNION ALL instead of plain UNION. With plain UNION, the engine will try to de-duplicate your result set. That is likely the source of your problem.
Secondly, you'll need indexes on owner in both tables, not just one. And, ideally, they'll be integer columns.
Thirdly, Randolph is right that you should not be using "*" in your SELECT statement. List out all the columns you want included. That is especially important in a UNION because the columns must match up exactly and, if there's a disagreement in the column order in your two tables you may be forcing some type conversion to go on that is costing you some time.
Finally, the phrase "There are two identical tables" is almost always a tip-off that your database is not optimally designed. These should probably be a single table. To indicate ownership of an item, your safety_dep_box table should contain only the ownerID and itemID of the item (to relate equipment and players), and possibly an additional autonumbered integer key column.

Answer 2

First off, don't use SELECT * in views ever. It's lazy code. Secondly, without knowing what the base tables look like, we're even less likely to be able to help you.
The reason it takes forever is because it has to build the full result and then filter it. You'll want indexes on your owner fields, whatever they may be.