Re: Is a SETLL faster than a CHAIN

Walden H. Leverich wrote:

Barbara,

I would imagine though, IF all you cared about was the existence check,
then SETLL would still be the winner since it only involves a trip to
the index, while chain requires a trip to the index, the table-space and
a load of the data. Make sense?

Yes, it does make sense, if the record is more likely to be found than not found. Although the SETLL also requires a reposition of the file cursor. But Tim's point in favour of CHAIN makes more sense, for the current case where the record is 99 times more likely to be not found. A SETLL would involve a reposition of the file even if the exact record was not found, unless the search argument was higher than all the records in the file, while a CHAIN would only require a trip to the index.

I suspect that the relative performance of SETLL and CHAIN is not determinable, or even guessable, through a thought-experiment. It might not even be determinable through a performance test, unless the test was very carefully crafted for a particular file, set of data, and set of search arguments.