Very interesting. Good experiment. For your next one, you can eliminate the advantage of the first and second links by randomizing the order each time you serve the page.

Yes, yes, a thousand times yes. Distributed systems that rely on all the pieces being up all the time are simply at odds with reality. Every interaction with someone else can result in a success, a failure, a rejection, or your request simply getting lost. Your design doesn't have to accept that fact, but failing to design for it doesn't make it go away. I wrote about this two years ago: When you build internet-scale distributed systems, you should always assume you are in flaky connection mode. Maybe the tubes are down today. Maybe your vendor’s server went down. Even with all the contracts and SLAs and angry phone calls in the world, you fundamentally don’t have any control over that box staying up and reachable when you need it.