How to remove multiple columns that have only one value across all rows in SQL Server 2008 R2

Question

I have a large table in SQL Server 2008 R2 and it contains many columns and millions of rows.

There are columns for which all rows have identical values.

For example

col1 col2 col3  col4 col5.....
a     b    c     1    null
a     d    e     1    null
a     f    g     1    null
a     h    I     1    null

I want to remove those columns such as col1, col4, col5 and more columns like those.

I do know drop column, I guess I just don't know how to select multiple columns in these case

How could I proceed in this case ?

Thanks very much

Answer 1

You can specify multiple columns in one statement:

alter table 
    mytable
drop column
    col4, col5

To see how many values a given column has, you can use count distinct :

Select
    count(distinct col1),
    count(distinct col2),
    count(distinct col3),
    ...
from
    mytable

Here's the outline of how to build the previous query dynamically:

Declare 
    @sql nvarchar(max) = N'select',
    @tab sysname = 'mytable', -- replace with table name
    @col sysname,
    @sep nvarchar(1)

Declare col_cursor cursor local fast_forward for
select
    name
from
    sys.columns
where
    object_id = object_id(@tab)

open col_cursor

fetch next from col_cursor into @col

while @@fetch_status = 0
begin
    set @sql += @sep + N' count(distinct ' + quotename(@col) + N') as ' 
        + quotename(@col)
    set @sep = N','
    fetch next from col_cursor into @col
end

close col_cursor
deallocate col_cursor

set @sql += ' from ' + quotename(@tab)

exec sp_executesql @sql