It sounds like your site needs to do a major analysis effort. If the jobs are I/O bound, you may be able to reduce timing by spreading workload across the channels evenly . If the jobs are CPU bound, you could look at running when the CPU is less busy. If neither of these ideas work, you r site may need to spend some money to upgrade (tape drives to use newest models as Dick suggested, or DASD to allow disk-to-disk copies instead of disk-to-tape, or perhaps a CPU upgrade). Since your pack backup times seem reasonable, there may not be much we can suggest -- only someone very familiar with your site could make suggestions based on analyzing the SMF data.