We were surprised to see that downtime peaked at 18.8% of the sites we monitor, compared to the 0.9% of the baseline.
What we can establish from our data is that 503 responses increased, but those don’t prove that someone shut down intentionally. Of course, all samples only cover the sites we monitor.
I’d love to get more data from other people, especially those who decided to stay online, or who entered maintenance mode while patches were being worked on.