what aspect of AWS is inconvenient here? IAM and VPC stuff just requires extra knowledge but once you understand it its not a hinderance that would get me shopping for a replacement.
This is kind of like the "why would I use Dropbox instead of cron+rsync" comment. There's lots things you have to do in AWS that have nothing to do with actually publishing your app. IAM, vpc, relevant vpc endpoints, secgroups/nacl, lambda, cloudwatch, cloudfront, API gateway, hopefully cloudformation/terraform, some wrapper for actually deploying the lambda, billing alerts, extra time because even though you know how to do this, it's easy to make trivial mistake that makes it not work.
Or... Just use vercel or similar packaged solution.
AWS has everything you could ever need, but that means you have a ton of trade offs to consider whenever you want to start something new.
I often want a one-off website deployed live on the internet to test something. For AWS I’m juggling in my head:
- Load Balancers
- API Gateway
- CloudFront
- S3 for static files
- Route 53
- Lambda config (or use Serverless to manage this or maybe ECS or EKS or maybe just spin up an EC2 instance…)
- Elastic beanstalk or Amplify instead of or in addition to the above?
- Cloudformation or Terraform to manage it? Or just build tech debt by doing it all manually?
- And how are we automating deployments? Maybe something custom in GitHub actions? I know AWS has some offerings here but they were cludgey last time I tried them.
Or with Vercel I:
- Type “vercel” and hit enter in the folder containing my web app.
They’re completely different worlds.
Sometimes I need the full power of AWS. Often I don’t.
all those things you described are a non issue if you use AWS Api gateway and Lambda. I'm struggling to understand why acknowledging these things is so much of a cognitive load for you.
In addition you can use Pulumi, Terraform, CDK but you could just do it from the web console and then generate those.
I don't know for me these are not enough of a pain for me to move off AWS.
Lemme put it this way. You know how people pay accountants to do their taxes or get a plumber to repair their sink? They can all be done theoretically by the user but it costs time to fix + time to learn about how to fix.
That's why people pay a premium for vercel(and other similar services like heroku) for hosting.
still not sold here. your analogy is misconstrued because those separate roles cannot be performed by one person in the first place whereas AWS is simply a matter of reading the documentation and building a knowledge of its architecture which compromises of many interconnected products. This is just how lot of enterprises function and AWS reflects this.
If you are starting out in AWS and you are overwhelmed its understandable but the if barrier to entry to Vercel is much lower you are not going to be differentiating yourself in the job market either.
The cost savings in the long run of having good AWS knowledge is always going to shine over whatever short term cognitive lift perceived by inexperienced AWS devs.
I would actually put Vercel in a lower tier of cloud practioners. For startups, that might be just what you need since there is a 99% chance you won't ever hit its unlimited tier limits (hyperbole).
In the time you've spent arguing about Vercel on here you could have actually tried it out and then you'd actually have your own opinion based on genuine experience.
I've used AWS Lambda professionally for years, Vercel is a totally different beast. I recently span-up a hobby project recently on Vercel rather than AWS Lambda (using Serverless Framework) to try out next.js and it took about 10 minutes to get it working whereas AWS Lambda would have taken a lot longer.
As an executive I need to carefully assess an offering by thinking of everything. Your reasoning is "shut up and use it". This is not how you convince c-suites.
If you've been really using Lambda on AWS and still struggling with IAM/VPC I would have to chalk this up as not amazon's fault
imagine forgoing these knowledge because you are lazy only to pay it out of your pocket when you realize vercel's pricing is significantly larger for your employers.
its not going to be fun when your startup runs out of money and you can't afford vercel anymore.
