Remote devices, such as sensors and communications devices, require continuously available power. In many applications, conventional approaches are too expensive, too large, or unreliable. For short-term needs, primary batteries may be used. However, they do not scale up well for long-term installations. Instead, energy harvesting methods must be used. Here, a system design approach is introduced that results in a highly reliable, highly available energy harvesting device for remote applications. First, a simulation method that uses climate data and target availability produces Pareto curves for energy storage and generation. This step determines the energy storage requirement in watt-hours and the energy generation requirement in watts. Cost, size, reliability, and longevity requirements are considered to choose particular storage and generation technologies, and then to specify particular components. The overall energy processing system is designed for modularity, fault tolerance, and energy flow control capability. Maximum power point tracking is used to optimize solar panel performance. The result is a highly reliable, highly available power source. Several prototypes have been constructed and tested. Experimental results are shown for one device that uses multicrystalline silicon solar cells and lithium-iron-phosphate batteries to achieve 100% availability. Future designers can use the same approach to design systems for a wide range of power requirements and installation locations.