Cloudflare’s latest Radar review is a useful reminder that “cloud reliability” is not only a question of whose dashboard is green. In the second quarter of 2026, the company logged a messy collection of Internet disruptions caused by storms, earthquakes, power failures, government shutdowns, cable damage, cloud-region trouble and even DNSSEC maintenance gone wrong.
The through-line for hosting customers is simple: the public Internet still has physical choke points and administrative choke points, and either can make a perfectly healthy website look broken from the outside.
The outages were not all “cloud outages” — but they looked like one to users
In its Q2 Internet disruption summary, Cloudflare highlighted Super Typhoon Sinlaku near Guam, earthquakes in northern Venezuela, a Tanzania power outage, repeated exam-related shutdowns in Sudan and Iraq, and a Saint Lucia cable cut. For site operators, those are not abstract network events: they translate into failed checkouts, unreachable dashboards, delayed support conversations and a familiar spike in “is the site down?” tickets.
The report also pointed to trouble around AWS’s Middle East region, noting that HTTP traffic to the UAE-based me-central-1 region remained low after earlier AWS service reports described physical damage affecting facilities in the region. That is the uncomfortable edge case in cloud planning: a region can be architected as software, sold as an availability target and still depend on buildings, fiber routes, power and geopolitical luck.
DNSSEC gets its own caution label
The most directly relevant item for domain and hosting teams was Germany’s .de DNSSEC incident. Cloudflare says a May 5 key rollover at DENIC began producing invalid signatures, causing validating resolvers to reject .de answers and return SERVFAIL until the issue was resolved. To end users, that did not look like cryptography. It looked like German websites, email and apps had suddenly stopped working.
That detail matters because DNSSEC is often treated as a checkbox: enable it, publish DS records, move on. The operational lesson is less comfortable. Registries, registrars, DNS hosts and site owners all sit in the blast radius when signing, rollover or validation behavior fails. Monitoring should include external DNS resolution from validating resolvers, not just origin uptime.
The Hosting Times view: Cloudflare’s report is not a reason to abandon DNSSEC, multi-region design or CDN fronting. It is a reason to test the boring parts. Keep registrar access current, document DNSSEC rollbacks, monitor from more than one geography, and make sure incident playbooks distinguish an origin outage from a resolver, route, cable or jurisdiction problem. The outage may not be yours, but customers will still arrive at your status page first.
Source: Cloudflare Blog / Cloudflare Radar.