索引电子邮件，mysql中的表中的电话列

Question

I have a table with above 5 million users with an index for three columns now.

I am updating specific column using email or phone, when am doing this sometimes i have to update 100k users, so to check with email or phone is taking a while to check first then get id and update based on id.

If I make email and phone indexed (both should have duplicated values), will my query execute better than now? Is it best to index email and phone as both saved as Strings in Database.

Answer 1

CREATE TABLE foo (
    ...
    PRIMARY KEY(id),
    INDEX(email),
    INDEX(phone)
)

If that hint does not suffice, please provide your current CREATE TABLE plus your attempts at SQL for the 'check' and 'update'.

If the question is about testing for two columns, here are some tips:

SELECT ...
    WHERE email = '...'
       OR phone = '...'

will be slow. In general OR does not optimize well.

This would be faster:

( SELECT id
    WHERE email = '...' )
UNION DISTINCT
( SELECT id
    WHERE phone = '...' )

INSERT ... ON DUPLICATE KEY UPDATE ... will not work because email and phone are not unique.

What is the business logic when there are two rows with a matching email? Do you update both? What about a email match, but phone mismatch?

If you need both to match, then this pair is optimal:

UNIQUE(email, phone)

INSERT ...
    ON DUPLICATE KEY UPDATE ...