Question

I have an SQL query that I am looking to optimize.

SELECT * 
FROM QUEUE_SMS_ALERT Q1
where ALERT_ORIGIN = "FOO" 
AND RECORD_ID is null 
and PHONE NOT IN (
    SELECT DISTINCT PHONE 
    FROM QUEUE_SMS_ALERT Q2 
    where Q2.ALERT_ORIGIN = "BAR"
);

Basically need to get all rows where ALERT_ORIGIN is “FOO” Which do not have a corresponding row in the same table with ALERT_ORIGIN “BAR”. The table contains abt 17000 rows and there are only abt 1000 records with ALERT_ORIGIN “BAR”. So my query is supposed to give me abt 16000 rows.

EDIT : The current query is very slow. I do not have any indexes currently.

Answer 1

I’m guessing that you have NULL values in the phone column which means NOT IN doesn’t work (so it’s “fix” not “optimise”). So I’ve written it with NOT EXISTS:

SELECT *
FROM QUEUE_SMS_ALERT Q1 
WHERE
    Q1.ALERT_ORIGIN = 'FOO'
    AND
    Q1.RECORD_ID is null
    AND 
    NOT EXISTS (SELECT *
        FROM QUEUE_SMS_ALERT Q2
        WHERE 
            Q2.ALERT_ORIGIN = 'BAR'
            AND
            Q1.PHONE = Q2.PHONE)

If it is slow rather than “wrong” then you need to use indexes. What do you have now?
For this query, you need an index on (ALERT_ORIGIN, PHONE, RECORD_ID).

Note: use single quotes for string delimiters