Search Results

For "thousands", splitting (or not) will make little difference to performance. Even for millions, it may not matter. For maintainability and performance, let's see the main queries. JSON may be an …

Get rid of user_addresses; instead, have an address_type (varchar) in the addresses table. That also gets rid of the over-normalized "address_types" table. Address_types might be "billing", "mail", " …

I care more about the performance than about NFness. Having a column for each option becomes unwieldy unless such additions are very rare. There is a third option -- the SET datatype allows up to 64 t …

Yes, but at most one can be AUTO_INCREMENT, else they would be fighting over how to number them. No, in the sense that is rarely wise to have two tables instead of one. There are exceptions; what is …

Turn things around. Table addresses needs a delivery_id. Ditto for the table storage. If you want to avoid having multiple rows for addresses or storages, then have two "many-to-many" tables to map b …

You don't need to stand on your head to get the id: Step 1: Populate Orders in a single statement: INSERT INTO Orders (scanId, customerId) -- Leave out; `id` it will be set SELECT DISTINCT scanI …

Normalization decreases the hassles; you decide where the cutoff is -- among speed/space/hassle/etc. …

Both of these schema anti-patterns have been repeatedly refuted: Do not have multiple 'identical' tables. Do not spread an array across columns. So have one table with lots of rows: sites …

A typical implementation of "multi-tenant" is to have a database for each client. Each database has the same few dozen tables. Security (if you care) is somewhat available at the database level. If y …

Perhaps a better policy: When registering a user, store the fee charged into the user's history. Derive the fee from some business logic that may or may not be stored in a database table. Note how …

Do NOT normalize any "continuous" or "numeric" values, such as a timestamp. First of all, TIMESTAMP and DATETIME each take 5 bytes. INT takes 4 bytes. So, that's not much savings. A 3-byte MEDIUMI …

In the table Chapter you should have a column book_id. That is the standard way to implement many:1. There no need for an extra table; that is only for many:many.

An "order" is composed of multiple "items", yes? Well, don't mix the two things in a single table. Similarly eBayOrders_account1 seems to be a mashup of orders, orderitems, users and payment. In gen …

Q1: "correct" in what sense - according to theory or practice or performance? Q2: JOINs are not that deadly. Q3: We need to see the SELECTs that you will use before we can judge your Schema. Q4: …

Which happens more often? Renaming/reordering versus fetching? If you are fetching most of the time, then make that more efficient (PK order), while letting the other be more costly. If you are ren …