I'm curious about how AWS Lambda compares to traditional back-end solutions. Can Lambda run various back-end operations like accessing databases, performing inferences, or sending forms? Additionally, can these functions be executed entirely from the front end? Is it feasible to completely rely on Lambda for back-end needs, or does this approach have inefficiencies? Also, how well does Lambda handle Node.js functionality? Can I use all Node.js functions with it?
4 Answers
Are you talking about 'puter'? If so, Lambda isn't a full server replacement—it's mainly a function runner. You will still need a proper server for things like WebSockets or maintaining persistent states.
Can Lambda act as a back-end? For many applications, yes! It's especially effective for simple tasks like API calls and email sending. But if you need long-running processes or real-time communication, a traditional Node server or a solution like ECS might be easier to handle. For side projects, you can use Vercel for frontend, Lambdas for backend functions, and a managed database like Supabase to simplify things.
In my experience, Lambda can't fully substitute a real server. Issues like cold starts and limited runtime can hinder performance for anything beyond small tasks. I moved a Node.js API to Lambda and noticed significant latency, so I kept a small EC2 instance for heavier tasks. Anyone else experiencing similar problems?
Yep, I inherited an app with 50 CRUD endpoints as individual Lambdas. We faced cold start issues too; it's manageable until a user is in the flow, then sudden 3-second delays hit hard. We ended up rebuilding our version for Kubernetes.
Yes, AWS Lambda can perform back-end tasks like accessing databases and sending emails, but whether you should is another matter. It can handle low-volume use well, but expect potential delays (cold starts) if traffic increases; that can also make it pricey.
Check out puter.js for an interesting take on this: https://developer.puter.com/.