The claim is that the service handles distribution across reliability, so I think the more interesting question is the odds of that _mechanism_ failing when your site would otherwise be up[1].

Similarly, 3-2-1 is a backup strategy and the pricing appears to already include multiple copies using the same mechanism so the correct calculation would be the cost of R2 plus whatever _different_ mechanism you choose for disaster recovery purposes such as on-premise storage or a completely different provider.

1. For example, if you use Cloudflare as your CDN / DNS provider and they have a catastrophic failure, the fact that your storage is inaccessible is just another facet of the same issue.

If you want to do proper 3-2-1, shouldn't you be storing the other backups on a different service?

That's the reason why most enterprises have 4-7 copies of their data..... No inherent geo-replication by default (and as CF shares, it's a "hotel California" problem.... Too expensive to egress completely from AWS.

What do you mean... The cloud does not need backups /s

Look up Bucket Replication Policies....

clearly the /s is not seen.....

that for those in the back is sarcasm....