Thanks for response.
For ssh and DBI we will have two separate programs, and will be easy to maintain. for running in sequential, we will have problem. Example: there is poosility of delay in process for few servers and might create problem when report won't be available on time.
I am thinking to use Thread::queue, will divide servers into 10-15 groups and kick off programm with GRP1..N as input parameter.
1. collector. pl GRP1 ## GRP1 will contain 30-40 dataserver
Please suggest if this approach is not efficient.
As I see it, you are currently neither clear of which has a higher priority, and what each task is to do. I really recommend that you separate the concerns. You can easily parallelize the task of information retrieval by starting programs in parallel. So do that instead of worrying about fork or threads.