Question

I have large table (~1,000,000 rows) that may contain duplicates values.

the table contain two columns (for example col a, col b) that together represent unique key, ID and last update date.

for example I can have table like:

id | a | b | update

1 | jon | smith | 1/1

2 | don | smith | 2/5

3 | bob | david | 1/1

4 | dan | lewis | 3/1

5 | bob | david | 3/1

As you can see for id 3 and 5 the table contain the same values in both a and b columns.
I would like to delete the rows that contain this kind of duplication , but keep the last updated row.

For this example I will have this table after deletion:
id | a | b | update

1 | jon | smith | 1/1

2 | don | smith | 2/5

4 | dan | lewis | 3/1

5 | bob | davis | 3/1

(id = 3 deleted ,since I already have a=bob and b=davis in row where id=5 and the update in this row is higher then the one in the deleted row)

Answer 1

delete from MyTable
where exists (
    select 1 from MyTable t2
    where MyTable.a=t2.a and MyTable.b=t2.b and MyTable.upd<t2.upd
)