Obviously, the first order of business (after determining that the interface that you are using is, in fact, thread-safe), is to determine that implementing this process using multiple threads actually will be faster. The only thing that multi-threading can realistically do is to overlap computation with I/O, and, to a certain but limited extent, to allow multiple I/O requests to be serviced in parallel. Everything, in your scenario, will depend upon the implementation capabilities of the Win32 subsystem with which you are ultimately communicating. If you, as you say, “know nothing about threads,” why do you have reason to believe that threads will actually help you? (I mean that as a perfectly serious question.)
Another often-overlooked strategy is simply to run multiple instances of the same Perl program on the same computer at the same time, arranging by some means for each of them to be issuing a different set of requests. Thus, with no added internal complexity to any one of the systems at all, as many copies as desired “of the whole thing” can be run in parallel.