I created Cloud Looking Glass as a CS grad research project to build a better status page for public cloud platforms — independent and based on what users actually experience (in my experience, cloud providers tend to underreport events on public status pages).
I modeled it as a time series problem: all tests are repeated every 5 minutes from three unrelated cloud accounts per platform (currently ~1,000 tests/min). The system provides full data plane, control plane, and networking coverage for AWS us-east-1, plus network testing across 25 other providers. Full coverage for GCP and Azure is coming soon.
It also has a cloud looking glass for real-time zonal, regional, cross-region, cross-cloud, and last-mile network testing (ping, tcping, traceroute, MTR, etc.).
Would love feedback, suggestions, or advice.
https://cloudlooking.glass#show=uptime