Question

I’m grabbing the last price from a specific product, but performance is really slow on over 3 million prices.

Does anyone have a better way of doing this? My server is getting hammered by this slow query.

prices.id is used to store the id of the store in the prices table so I can join it with the stores.id from the stores table.

SELECT prices.id, prices.price, prices.timelog, prices.user_id FROM prices
WHERE prices.id IN (SELECT stores.id FROM stores WHERE city = "miami" )
AND prices.product_id = 1
AND prices.timelog IN 
  (SELECT MAX( lastprice.timelog ) 
   FROM prices AS lastprice 
   WHERE lastprice.id = prices.id AND lastprice.product_id = 1)

Answer 1

3 million prices might mean your application could benefit from some denormalisation. (i.e. flag the last price in some manner, remember: write-once, read-many is often worth the overhead of a slower write).

With the current data, worth a try are:

SELECT p1.id, p1.price, p1.timelog, p1.user_id 
FROM prices p1
JOIN stores 
ON stores.id = prices.id
AND stores.city = "miami"
LEFT JOIN prices p2
ON p2.product_id = p1.product_id
AND p2.id = p1.id
AND p2.timelog > p1.timelog
WHERE p1.product_id = 1 AND p2.id IS NULL;

… and another a bit faster option which always escapes me this time of night 🙂

Keep in mind a lot can change with the proper indexes on tables, so by all means, run an EXPLAIN on the lot.