The potential availability of storage space at cellular and femtocell base-stations (BSs) raises the following question: How should one optimize performance through both load balancing and content replication when requests can be sent to several such BSs? We formally introduce an optimization model to address this question and propose an online algorithm for dynamic caching and request assignment. Crucially our request assignment scheme is based on a server price signal that jointly refects content and bandwidth availability. We prove that our algorithm is optimal and stable in a limiting regime that is obtained by scaling the arrival rates and content chunking. From an implementation standpoint, guided by the online algorithm we design a light-weight scheme for request assignments that is based on load and cache-miss cost signals; for cache replacements, we propose to use the popular LRU (Least Recently Used) strategy. Through simulations, we exhibit the efficacy of our joint-price based request assignment strategy in comparison to the common practices of assigning requests purely based on either bandwidth availability or content availability.