Roberta Arcoverde, Director at Stack Overflow, previously a staff engineer and tech lead:
Stack Overflow platform:
* .. backed by just 9 servers,
* .. run on-prem from own data center (not cloud),
* .. team of 50 engineers (up from 10 in 2014),
* .. deployed as monolith in 4min multiple times daily,
* .. pragmatic, ignore trends like k8s and microservices.
> Giving SQL 1.5 TB of RAM was more effective than caching page fragments in Redis.
https://hanselminutes.com/847/engineering-stack-overflow-with-roberta-arcoverde
𝚜𝚎𝚕𝚎𝚊 
VERY happy that they dont run shit "in the cloud"
@krinkle meanwhile job requirements for open positions at stack overflow: "We expect to see: Prior experience successfully migrating infrastructure and systems to a cloud environment (Azure, AWS, or Google)."
@IngaLovinde She does say that they're constantly reevaluating their choices and some bits and pieces are using cloud services already.
At the same time everything is currently optimized for low latency and direct communication with the database, like writing SQL queries was being mentioned?
When you move that to any cloud, latencies are added everywhere (including inside the database), and suddenly you need caching layers and add complexity to have that managed and synchronized...
@krinkle thanks for sharing!
@krinkle With transcript available here for those who prefer to read : https://app.podscribe.ai/episode/83433649