Regarding "Duplicate Record WRITE Performance" (TechTalk, MC, January 1995): while using a CHAIN before a WRITE can outperform a conditioned WRITE operation when writing duplicate records to a unique file, I have some information you may find useful. My company tested this and found that, if the number of duplicates is small (specifically less than 12 percent from the primary input file), it is more efficient to do the conditioned WRITE than suffer the overhead of the CHAIN operation. This test was done about a year ago on a B50 under V2R1, but I am sure current results would be comparable.

- Jonathan Frankel