In a scenario where you need to design a system that controls for cost and latency SLAs for production streaming jobs, how would you approach the design to ensure that both cost and latency are optimized? Discuss the strategies you would employ to balance these two critical factors.