Basic items to keep in mind:
1 - dumps/loads that do not use compression are solely disk intensive
2 - dumps/loads that use compression are cpu and disk intensive, with higher compression levels typically requiring more cpu/time to complete the dump/load operations
3 - dumps will put a load on the disk controller(s) managing the database devices in the form of reads from the database devices; dumps will also put a load on the disk controller(s) managing the dump devices in the form of writes to the dump devices (less so for compressed dumps).
4 - loads will put a load on the disk controller(s) managing the database devices in the form of writes to the database devices; loads will also put a load on the disk controller(s) managing the dump devices in the form of reads from the dump devices (less so for compressed dumps).
5 - striped operations are typically used to speed up dump/load operations by using more system resources (eg, each stripe will be assigned a pair of sybmultbuf processes to perform the necessary disk IOs; the stripes' sybmultbuf processes will generate large volumes of IO requests in a shorter period of time that would be seen with a single-striped operation)
6 - striped *compression* operations will increase the load on X cpus where 'X' is the number of stripes (ie, one of the sybmutlbuf processes assigned to each stripe will be used for the cpu-intensive (un)compression operations)
So in answer to your questions ... "It depends"
a - if your disks and/or controllers are already running at/near max throughput then any dump/load operations will likely cause an increase in IO service times thus slowing down anything that relies on disk IOs
b - if your system is already running at/near max cpu levels then any dump/load operations that use compression could bog down all operations that rely on cpu resources
c - striped operations are typically used to speed up dump/load operations by performing more disk IOs in parallel; obviously this is only of benefit if your disk subsystem can handle the increased volume of IO requests with minimal impact to service times
d - striped *compressed* operations will put a load on X cpus, which should only be attempted if you have the available cpu resources otherwise you will likely starve all cpu-intensive operations on the machine
It's up to the DBA (and OS/disk admins) to monitor the various machines in the environment to see where/when striped (compressed) operations can be performed without adversely affecting other (more important?) processes running on the same machine (and/or disk subsystem). You may find that some of your systems can support striped/uncompressed operations, while others can support striped/compressed operations, while others may need to stick with single-stripe/uncompressed operations, etc.
NOTE: If you're using SAN-based disk subsystems you may also need to work with the network/SAN folks to insure your striped operations do not adversely affect other systems sitting on the same network segments and/or SAN slices.
NOTE: If you're running ASE in a virtual environment then large volumes of compressed dumps/loads will eat up cpu resources on the underlying hardware, which in turn could adversely affect other processes running on the same hardware (eg, the physical resources are over-subscribed by a large number of normally lightly-loaded VMs).