Raspberry Pi 4 launch site runs on a Pi 4 cluster

previously: https://news.ycombinator.com/item?id=20265751

Ars' article adds the performance comparison, but it's interesting that the main source [1] suggests they're using power-over-ethernet to power these. (At least, they were for the previous set.)

https://blog.mythic-beasts.com/2019/06/22/raspberry-pi-on-ra...

The photo shows a PoE switch.

I guess this is why the site was down the morning of release. It didn't have proper autoscaling.

The article describes the downtime as being caused by a "severe BGP error Verizon made".

Full story on the outage: https://www.theregister.co.uk/2019/06/24/verizon_bgp_misconf...

A BGP error could be completely out of their control and nothing to do with using a Pi at all.