Given two approaches to "what if" scenarios:

1) Do work in a transaction. Involves a fairly complex process, many updates. Evaluate results, and if undesireable, rollback and try with new parameters, else commit. The try/rollback may be repeated several times until a desireable result is achieved.

2) Create a workspace, with all the necessary tables, create a savepoint, and perform the work. If results are undesireable, rollback to the savepoint and try again with other parameters, else merge results back into the "live" tables.

Currently I am doing option 1. The process involved is scheduling a series of events over time. There are sequencing rules and other constraints, and the parameters that can be adjusted are the relative priorities of the rules. Desireability of the result is determined by the number of rules violated, weighted by the priority of the rule. I have several differnt rule sets, which I try in sequence until a good result is obtained.

Is there enough information here to say that it is likely that a workspace rollback to a savepoint would be faster than a transaction rollback, for a "large" rollback (tens or hundreds of thousands of rows)? I really just discovered workspace manager, the ability to have the "what if" results contained in a workspace is appealing. But if putting this process into a workspace is likely (or certainly) going to perform worse, I don't want to go that route, at least not in an attempt to improve performance.