An improved scalability testing architecture for distributed applications
allows testing of scalability with respect to a given number of computers
using a test network having only a small fraction of that number of
computers. Each client computer in the test network hosts multiple
bootable systems in bootable partitions which timeshare the host machine
on a mutually exclusive basis. A launcher in each bootable system
verifies scheduling information for an instance of an application under
test residing in the same partition and either permits or prevents
continued operation of the instance based on the scheduling information.
If continued operation is to be prevented, the launcher triggers a boot
next routine to exit and cede control of the computer to the next
bootable system on the computer.