Reliability and availability in microservice architectures should be measured differently than those in traditional monolithic systems. The interrelations and fluidity among microservices tend to be ignored by the classic metrics. We propose the Service Availability Ratio (SAR), which calculates service availability from end-to-end, taking into account direct as well as indirect dependencies on services through invocation chain logs. SAR gives a holistic picture of system availability and pinpoints weaknesses in reliability for better design and testing of systems. This paper will describe how SAR is automatically extracted and computed, considering challenges that arise from distribution and dynamism inherent in microservices environments.