In recent interviews, I asked candidates a system-design question about managing failures in a serverless, event-driven architecture. I was surprised by how many didn't include retry mechanisms or a D

Markdown supported · ⌘↵ to post
No responses yet.