If it's just a grand total, then the data step may save you a little,
but not large amounts of time. As soon as grouping is involved,
however, SQL can cost you a lot, especially to the extent that it has to
do unnecessary grouping.
Can you give us a little more information about the problem?
> Hello all
> I have to sum up variables in a dataset that has some 30 million observations.
> Right now we are using proc sql to do it. But, it is taking a lot of time. How
> Can I improve the performance? Can I use data step with retain? Are there any
> other solutions?
> Any suggestions?
> Any help is greatly appreciated.