Data-Driven Insights & Real-Time Testing of Cloud Availability & Performance

Cloud Looking Glass is a research project focused on empirically analyzing availability and performance of hyperscale cloud services. Using triple-redundant accounts per provider and with test agents deployed across regions and zones, we continuously measure key functional characteristics, including control plane response times for cloud service CRUD operations (e.g., provision compute, storage, serverless, databases), network and storage latency (data plane), and many others. Failures are recorded automatically and manually verified.