Very few computer systems that have been deployed in rural developing regions manage to stay operationally sustainable over the long term; most systems do not go beyond the pilot phase. The reasons for this failure vary: components fail often due to poor power quality, fault diagnosis is hard to achieve in the absence of local expertise and reliable connectivity for remote experts, and fault prediction is non-existent. Any solution addressing these issues must be extremely low-cost for rural viability. We take a broad systemic view of the problem, document the operational challenges in detail, and present low-cost and sustainable solutions for several aspects of the system including monitoring, power, backchannels, recovery mechanisms, and software. Our work in the last three years has led to the deployment and scaling of two rural wireless networks: (1) the Aravind telemedicine network in southern India supports videoconferencing for 3000 rural patients per month, and is targeting 500,000 patient examinations per year, and (2) the AirJaldi network in nothern India provides Internet access and VoIP services to 10,000 rural users.