Sometimes, rather than a “hot failover” multi-region approach, sometimes a more sensible fallback degraded experience might make sense.
This tweet is unavailable
1
10
27
I remember at the Guardian they would regularly crawl the website. If the main system went down, a static site could be switched to easily in its place.
1
1
8
Or perhaps an e-commerce company with a simple static version of the website which doesn’t know what’s in stock, but let’s you place orders to be processed later when the website is back.
5
1
15
Replying to @samnewman
Reminds me of one system I reviewed that had a separated, decentralized subsystem called “ordertaker”. Even when everything else broke down, this was the one thing that kept running no matter what

Mar 26, 2021 · 7:22 AM UTC
1
2