System provisioning is the problem of determining the number of resources required to accomplish a complicated system level task, e.g. tracking or discriminating between N targets. This is a central problem in multi-target tracking with synthetic aperture radars where the number of targets can easily exceed the available resources. This paper treats the following conservative sensor provisioning problem: dynamically assign R platforms to process N moving targets in a way that guarantees that the radar maintains track on all targets. We propose a solution to this problem that guarantees a prescribed level of system performance, e.g., multiple target detection and position uncertainty levels, regardless of the scenario. The operational context of the paper is computational provisioning in synthetic aperture radar (SAR) that dynamically assigns different computers to tracking different targets.