Optical phased arrays (OPAs) are a promising technology for the realization of fast and compact non-mechanical optical beam steering. While many experimental demonstrations of integrated OPAs exist in the literature, it is challenging to evaluate their suitability for real-world applications due to the lack of system-level performance requirements. Here, we derive such performance requirements for two of the most promising OPA applications - namely free space optical communications (FSOC) and light detection and ranging (LIDAR) - and show that traditional uniformly spaced OPA architectures likely cannot reach the required performance. In response, we propose the use of non-uniformly spaced OPAs, analyze its performance tradeoffs and show that in certain scenarios they can offer superior performance with decreased complexity.