Bad results may contain very information. So I can help you, tell me what is the experimental design used. For example: number of VMs, workloads, and the scenario considered.

Best Regards,

i am working in allocating cloudlets to vms i use pso GA for this purpose when using 5 vm give me good results but when i change it give me bad results 

the good result is minimizing the execution time than the default algorithm and bad result give me more time than the default one 

thanks for your help

What are you changing? The VMs count? The algorithm and parameters are the same?Can you give me an example in numbers?

Best,

the number of vms

The number of VMS influence the performance, especially if they are co-hosted on a single physical host. This occurs due to interference of virtualization. In this case, the VM is equal workload, consequently, they compete for the same resources.

That depends on how you are simulanto these VMs and the resources involved.

Best, 

Carlos

this right if when increasing the number of vms but when increase or decrease it from specific number the results become very bad 

Have you checked the limits of the resources of the VM and the machine running the simulation? The parameters are correct?