Abstract
We report here a recent case on the construction of a fault-tolerant information infrastructure for a university hospital with 200 beds, which plays a central role in regional medicine. The risk assessment was conducted to prevent failure of electric supply according to business continuity plan. The hospital is provided redundant power from two power substations, and an in-house generator is connected to the power-supply circuit. Servers and storages except for the nurse scheduling server and a few servers for system development are connected to the generator-equipped circuit via UPS, and shut down at the outage of the circuit to keep the integrity of stored data. Power outage-test revealed that the systems worked and stopped according to the plan. Images and waveforms generated in the hospital are backed up to a DICOM repository and an MFER repository prepared in another hospital of the university through triplicate VPN lines with carrier diversity for disaster recovery. The use of the on-premise cloud minimized servers and storages, and reduced maximum power consumption from 80KvA to 52KvA. Airspace of the server room is separated into cold aisle and hot aisle, and a simple outdoor air cooling mechanism using a combination of hot-air exhaust system with fresh air intake system is installed. If the outside air temperature decreased from 18℃ to around 0℃, the power consumption of the cooling compressors of the air conditioners decreased to less than 50% by the cooling mechanism.
