Cloudflare shares some insights from their redesign of their KV-storage following the Google's outage.

In nutshell, they use a combination of their distributed database and a vendor's cloud storage depending on the size of data. This line I find particularly interesting:


Cloudflare engineers explain that "for workloads dominated by sub-1KB objects at this scale, database storage becomes significantly more efficient and cost-effective than traditional object storage."

#cdn #databases