I just did a new test using a cursor this time.

CREATE table #regions (

 ipFloat float null,

 countryId int null)

insert into #regions (ipFloat) select distinct top 500 ipFloat from tbl_logs

declare @i float, @id int

DECLARE cSE CURSOR READ_ONLY  

FOR select ipFLoat from #regions

OPEN cSE

FETCH NEXT FROM cSE INTO @i

WHILE @@FETCH_STATUS = 0

BEGIN

 select @id = countryId from tbl_ip2countries where @i between ipFrom and ipTo

 update #regions set countryId = @id where ipFloat = @i

FETCH NEXT FROM cSE INTO @i

END

CLOSE cSE

DEALLOCATE cSE

select * from #regions

drop table #regions

The results are quite surprising! For 100 ips, it took only 6 seconds instead of minutes.

But I'm sure there's a way to get this done quickly whitout using a cursor. I just can't find it out.

Thanks for any idea,

Stephane