[scarface74]

As Werner Vogels says “everything fails all the time.” How long can Amazon Retail/AWS, Google/GCP/YouTube, or Microsoft survive without continuously replacing hardware? How much degradation of their service can they withstand as they are unable to replace hardware?

If hard drives fail at YouTube and they can’t replace them and keep up with demand, does YouTube start deleting old videos to make room for new ones? Does AWS stop promising 6 way redundancy across three data centers? When a server goes south in us-east-1 and they don’t have a replacement, then what?

[fomine3]

In emergency situation, I can come up with mitigation something like: YouTube could reduce available video quality options only for 144p/480p/1080p/2160p for some rarely watched videos and remove other quality converted videos (but keep original file internally). Such mitigation will work for consumer oriented services, but not work for AWS/GCP. Possibly they can also stop internal analysis/research project that consumes a lot of resources.

They can also delay replacing hardware by accepting higher power usage and failure rate.

Yes it's serious problem but not critical as much as hardware company for immediately.