Sanity check that you’ve got the same plan in both circumstances. Remember
that sql*plus autotrace gives you an explain plan not an execution plan.
If that’s the same then enable extended SQL trace for your loadrunner
session, check out exactly where these differences come from. The row
source stats part of the tkprof report should help you see which line of
the plan is responsible, the stats should also give you decent clues.
If you are dealing with bind variables then make sure you are reporting the
ones used in your trace files. Are you testing with the same values? Or is
there a chance that your loadrunner session is using high selectivity
values but your sql*plus is not?
Hope that helps you get started in the right direction,
Andrew
