The table in question has ~30mio records. Using Entity Framework I write a LINQ Query like this:
dbContext.MyTable.FirstOrDefault(t => t.Col3 == "BQJCRHHNABKAKU-KBQPJGBKSA-N");
Devart DotConnect for Oracle generates this:
SELECT
Extent1.COL1,
Extent1.COL2,
Extent1.COL3
FROM MY_TABLE Extent1
WHERE (Extent1.COL3 = :p__linq__0) OR ((Extent1.COL3 IS NULL) AND (:p__linq__0 IS NULL))
FETCH FIRST 1 ROWS ONLY
The query takes about four minutes, obviously a full table scan.
However, handcrafting this SQL:
SELECT
Extent1.COL1,
Extent1.COL2,
Extent1.COL3
FROM MY_TABLE Extent1
WHERE Extent1.COL3 = :p__linq__0
FETCH FIRST 1 ROWS ONLY
returns the expected match in 200ms.
Question: Why is it so? I would expect the query optimizer to note that the right part is false if the parameter is not null, so why doesn't the first query hit the index?
Please set UseCSharpNullComparisonBehavior=false explicitly:
If this doesn't help, send us a small test project with the corresponding DDL script so that we can investigate the issue.