Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing availability metrics for an integrated service. One of the methods includes generating, for a service installed on a software platform, a plurality of availability logs, each availability log representing an occurrence of the service becoming unavailable. The plurality of availability logs are aggregated according to one or more aggregation criteria. The aggregated availability logs are processed to compute one or more availability metrics for the service, wherein each availability metric quantifies the availability of the service to process requests from the plurality of workloads in the presence of system failures and interruptions. An availability rating for the service is computed from the one or more availability metrics.