An approach for providing a user with resumption data before pausing or terminating a partially-processed job that is processing in a distributed system. Certain applications, when paused, are not able to resume processing data at precisely the point where it was paused. As a result, users surrender some, or all, of the processed portion of the data when processing is paused. In one embodiment of the invention, a user is provided with resumption data before the user confirms a request to pause the processing. Resumption data from each of the service nodes that is processing a segment of source data are collected and aggregated. Resumption data reflects a measure of processed data that would require re-processing if the processing were paused and resumed. Alternatively, resumption data includes a duration of spent processing time that would be lost if the processing were paused and resumed.