It's a commonly-repeated comment that early stage startups should avoid K8s at all cost. As someone who had to manage it on a baremetal infrastructure in the past, I get where that comes from - Kubernetes has been historically hard to setup, you'd need to spend a lot of time learning the concepts and how to write the YAML configs, etc.
However, hosted K8s options have improved significantly in recent years (all cloud providers have Kubernetes options that are pretty much self-managed), and I feel like with LLMs, it's become extremely easy to read & write deployment configs.
What's your thoughts about adopting K8s as infrastructure early on (say, when you have initial customer fit and a team of 5+ engineers) and standardizing around it? How early is too early? What pitfalls do you think still exist today?
Comments URL: https://news.ycombinator.com/item?id=44976292
Points: 8
# Comments: 2
Login to add comment
Other posts in this group

Article URL: https://gwern.net/everything
Comments URL: https://news.ycombinator.com/item?
Hey HN! I’m Lukas Dickie the founder of Changefly and I’m truly excited to share with you our latest release of Changefly ID with Anonymized Identity & Age Verification.
By putting privacy first
Article URL: https://demo.dynamicslab.ai/chaos
Comments URL: https://news.ycombinator
I recently worked on running a thorough healthcare eval on GPT-5. The results show a (slight) regression in GPT-5 performance compared to GPT-4 era models.
I found this to be an interesting find