Combining mobile edge computing (MEC), the multi-drone aided space-air-ground integrated Internet of things (SAG-IoT) networks can provide ground IoT devices (GIDs) highquality wireless access and computing services. However, the diverse tasks, moving drones, and limited network resources reveal great challenges for the task offloading and resource allocation scheme exploitation. Especially, given the restricted computation resources, how to make full use of available applications deployed on MEC servers (MECSs) to compute various types of tasks, is even an important issue. To the best of our knowledge, it is an entirely new problem since most existing works in this line assume that all types of applications can be deployed on one MECS so as to process various offloaded tasks. Toward this end, we present this paper to investigate inter-server computation offloading, resource allocation, and drone deployment to minimize the overall computation overhead of all GIDs. An iteratively optimization algorithm is proposed which alternately utilizes heuristic greedy and successive convex approximation methods. Simulation results verify that, for different GID numbers, optimization schemes, and computing models, our devised schemes can not only significantly reduce the overall computation overhead but also achieve optimal decisions of computation offloading, spectrum allocation, and drone deployment.