How is COUNT(*) different to COUNT(id)?

Asked Aug 20 '17 at 09:26

Active Aug 20 '17 at 09:26

Viewed 52 times

COUNT(id): query cost: 463

COUNT(*): query cost: 91.47

Can someone help me to understand why COUNT(id) is more expensive than COUNT(*)?

asked Aug 20 '17 at 09:26

Tony Dinh

3

`Count(*) counts all rows` , And `COUNT(column) counts non-NULLs only ` and have to access every row to see if the value is null or not See this post [count(*) vs count(column-name)](https://stackoverflow.com/questions/3003457/count-vs-countcolumn-name-which-is-more-correct) – Yagami Light Aug 20 '17 at 09:30
What is the cost if you have a "covering index": `INDEX(song_id, id)` – Rick James Aug 21 '17 at 18:30

0 Answers0