how can you count up the number of columns in a row that have a certain value

we have a table with 25 million rows. each row has

acct_# meter_# date read1 read2........read24

1324 123 01/01/99 14 0 4

we would like a count of the rows that have more than 16 of the 24 read columns with a 0 value. we need to exclude then from our analysis.

can somebody post SQL or pl/sql that does this??

also would a bitmap index on the date be of use in speeding up searchs by date they go from 7/1/1998 to 12/1/1998, 180 different dates each repeated 250,000 times or would a regular one better? Received on Thu Mar 25 1999 - 18:34:03 CST