Dropbox is not a relevant example to the original comment IMO - they clearly need cheap storage above all else; it should be no surprise that someone can self-host something so single-minded as a storage application without all the cloud provider baubles.

Dropbox makes sense. They're an egress-heavy business. Other than media companies (Netflix, Disney Plus, et al.) there aren't that many egress-heavy businesses, so AWS continues to make sense.

Is this true?

I thought dropbox moved their 30PB+ data lake ONTO aws to get off of Hadoop or something because trying to do this on-prem, even with tons of tech talent and money, was not working.

They complained about onprem requiring 3 YEAR forecasts for capacity planning given their scale.

Here is what they said in 2020 for benefits of AWS:

---------------------

Hosts 40 PB of analytics data and supports 1 PB of data growth a month Optimizes costs by moving cold data to Amazon S3 Glacier Deep Archive Uses Amazon EC2 Spot Instances for 15–50% of compute capacity Doubles compute footprint using Amazon EC2 Spot Instances Enables the testing of new technologies without damaging data or affecting users Improved performance by six times for some job types Deletes hundreds of files in a few seconds compared to 30–40 minutes Runs more than 100,000 analytics jobs and tens of thousands of one-time jobs daily

---

https://www.youtube.com/watch?v=6x-XGJQwk2M

Maybe this has changed since 2020