根据第一个表合并来自多个表的数据

Question

I have a table full of items, at the moment there is only one type and everything about that item is in the same table. Now I want to add two more types. A number of columns (>10) will be unique to each type, so I don't just want to have 30 columns in the table when 20 of them will be null values, so I'm thinking of splitting them up:

tbl_items_common
tbl_items_type1 [for data unique to this type]
tbl_items_type2 [for data unique to this type]
tbl_items_type3 [for data unique to this type]

All the tables will have a common field 'id', for matching up the items and data.

Now I'm thinking about the coding and thinking:

Should I bother splitting these up at all, or should I just have one big table. Makes inserting and querying easier, but the db is a bit more 'messy'.

Splitting the data up makes the db a bit more isolated, but makes code more complicated.

Or, I could just use 3 completely separate tables with the some of the columns the same between them, but then I'm not sure how to generate a unique id that is unique over all 3 tables.

If I do decide on splitting the data up, is it possible to get all the data in one query (embedded case/select maybe?), or would I have to do a SELECT first to find the item type and then another to grab the rest of the data? Not pretty when it comes to reporting.

So what do I want to know? (1) Which method would you choose and (2) If you choose the second, how would you query that data?

Answer 1

You have a few options. Personally, I'd keep all the important (and similar) data for your items in one table, and have 3 columns containing detail IDs, you could then create a couple of different tables containing the extra data for each type and JOIN that data to each query.

Another option is you could have three separate tables containing items, but I don't advise that as it's not good database design practice. If you did want to do that though, you could either select from all three tables in one query: SELECT * FROM table1, table2, table3

... or manually match the column count using null values and perform a UNION query:

SELECT f1, f2, f3, f4, null as f5 FROM table1
UNION ALL
SELECT f1, f2, f2, null as f4, null as f5 FROM table2
UNION ALL
SELECT f1, f2, f3, f4, f5 FROM table3 -- table 3 contains 5 columns as an example

... but that could get messy as well if you have lots of columns unique to each type. I'd suggest going with three separate tables for each item type's details :

SELECT * FROM items I
LEFT JOIN item_type1 I1 ON I.item1_detail_id = I1.id
LEFT JOIN item_type2 I2 ON I.item2_detail_id = I2.id
LEFT JOIN item_type3 I3 ON I.item3_detail_id = I3.id