Serverless has benefits. Each execution is isolated, has built in logs (cloudwatch if on aws), and auto scales based on traffic. Its also pretty cheap compared to running a server 24/7. All of these benefits can allow developers to focus on implementing business requirements instead of trying to architect a scalable system by themselves.
It’s very easy to write a node server with one endpoint that accidentally crashes your entire node application, which means all your requests will be dead until your server restarts. When your serverless function crashes, it only affects one request.
Serverless has benefits. Each execution is isolated, has built in logs (cloudwatch if on aws), and auto scales based on traffic. Its also pretty cheap compared to running a server 24/7. All of these benefits can allow developers to focus on implementing business requirements instead of trying to architect a scalable system by themselves.
It’s very easy to write a node server with one endpoint that accidentally crashes your entire node application, which means all your requests will be dead until your server restarts. When your serverless function crashes, it only affects one request.
In fact, we just need a generalized and standardized server scaling method. But maybe this is a phase non-negligible.