Understanding the nature of commonsense reasoning is one of the deepest questions of cognitive science. Prior work has proposed analogy as a mechanism for commonsense reasoning, with prior simulations focusing on reasoning about continuous behavior of physical systems. This paper examines how analogy might be used in commonsense more broadly. The two contributions are (1) the idea of common sense units, intermediate-sized collections of facts extracted from experience (including cultural experience) which improves analogical retrieval and simplifies inferencing, and (2) analogical chaining, where multiple rounds of analogical retrieval and mapping are used to rapidly construct explanations and predictions. We illustrate these ideas via an implemented computational model, tested on examples from an independently-developed test of commonsense reasoning.