Check for multiple columns with Single Value using SELECT query - sql

Question

I have a table fruits. Now in UI, I provide a single field for a search criteria. Using this single Search criteria, i want to search in multiple columns of Fruits Table.
Consider Fruits table contains Columns ID,Desc,Price,Quant,Stock. Here Price,Quant are integers and Stock is a varchar.
I have tried the below query which returns the results, but i am worried about the performance.
Suppose assume user enters 2 in the field provided in UI and clicks on search then query will be as shown below
select ID, Desc, Price, Quant, Stock
from Fruits
where Price = '2'
or Quant = '2'
or stock = '2'
Is this the right way to search for multiple columns of same table? Also will be any effect on performance?

Answer 1

First, you want to be sure that the types are compatible. In all likelihood these values are numbers, so drop the quotes:
select ID, Desc, Price, Quant, Stock
from Fruits f
where Price = 2 or Quant = 2 or stock = 2;
This can more simply be written as:
select ID, Desc, Price, Quant, Stock
from Fruits f
where 2 in (Price, Quant, Stock);
but that will not help performance.
In most databases your query will require a full table scan -- although some databases support a particular type of index scan called a skip scan which can help.
The only way I can think to get around that is to have a separate index on each column:
create index idx_fruits_price on fruits(price);
create index idx_fruits_quant on fruits(quant, price);
create index idx_fruits_stock on fruits(stock, quant, price);
(You'll see why the extra columns are helpful.)
And then use union all:
select ID, Desc, Price, Quant, Stock
from Fruits f
where Price = 2
union all
select ID, Desc, Price, Quant, Stock
from Fruits f
where quant = 2 and price <> 2
union all
select ID, Desc, Price, Quant, Stock
from Fruits f
where stock = 2 and price <> 2 and stock <> 2;
Each of the subqueries can use one of the indexes. Because of the inequalities, the results are exclusive -- assuming the column values are not null. If nulls are allowed, the logic can be adjusted to handle that.