from Hacker News

Hidden costs of self-hosted front ends

by sebnun on 7/27/24, 2:42 PM with 2 comments

  • by compootr on 7/27/24, 3:07 PM

    Of course, coming from vercel this is kinda disconnected from reality (and quite a bit AI-written) but:

    Tightly coupled frontends, or frontends that exist within the same codebase or deployment unit as your backend architecture, limit your ability to adapt.

    > The interior of each company’s container is unique, meaning that onboarding developers is tricky and most new features must be built from scratch and continuously maintained.

    I agree it's unique, but for "new features must be built"? If in software, that's how that works, and for something in linux, you install a pacjage in your build step

    > Maintaining a healthy development lifecycle is a fully manual, resource-intensive process. As traffic fluctuates, teams must provision the appropriate amount of infrastructure to adapt, or else face downtime or security breaches.

    Heard of CI, k8s and autoscaling?

    > Self-hosted setups struggle to optimally serve dynamic data to customers halfway around the world.

    This can be true, but just purchase PoPs as your business needs scale. If you're some small startup you don't need 9 9's of uptime.

  • by janmarsal on 7/27/24, 3:27 PM

    >Self-hosted apps come in many flavors, but they are often containerized, orchestrated with a tool like Kubernetes, and deployed on a cloud provider like GCP, Azure, or AWS.

    I thought self-hosting is the opposite of using cloud providers.