Many companies use AWS services because they’re trendy, and then they get very high bills.

AWS Lambda, Cognito, even CloudFront…

What they don’t know is that they could be saving more than 70% on those resources if they understand how AWS got started in the world of cloud servers, which is still the foundation of all the expensive “serverless” services that are trendy nowadays:

The core base stack is:

• AWS EC2 (web server or VPS)

• AWS EBS – storage (it comes with EC2, but it’s such a good service that it deserves to be mentioned separately)

That’s it, nothing more. Just a typical web server with its disk storage system.

You can practically rent a fleet of web servers, or just one, and host there the database (PostgreSQL for example), the media storage (images, files, documents, etc.), and the web application itself (Next.js, PHP, Express.js, Python, etc.).

There’s no incompatibility: it’s a web server on Linux, it can host anything.

Of course, if you want to scale or get better organized, you can put each thing on a separate web server: database separate, media disk handler separate, web app separate, and a build server separate (in case you don’t want to keep dealing with GitHub’s inconsistencies).

You can even replace AWS S3 if you don’t need many of its advanced features: just attach an EBS volume to the web server where your application runs, exclusively for handling media files; you can choose less “powerful” models that are much cheaper and reach almost the same price level as AWS S3: instead of GP3, choose HDD disks (io1 volumes) or SSD (io2 volumes), etc. And install a media handler library like MinIO, or use your own web app’s media handler. Optimize and compress the images so they’re lightweight.

CDN? CloudFront isn’t necessary; just implement Cloudflare CDN: it’s free, it has unlimited bandwidth, and it has very good performance for most web applications and software. Unless you need hundreds of TB of data output per month, Cloudflare remains a very solid option instead of using CloudFront, which charges for every GB of data output per month.

Another option is BunnyCDN, which gives you full control over how many TB you want to spend per month and has no surprise bills.

What happens if you have a lot of traffic, handle a gigantic amount of data, or need to process a lot of computing per second in your web app?

AWS EC2 has ways to solve this, either with AWS’s own services or manually (remember, these are Linux web servers):

• Set up a public web server, install nginx and turn it into a load balancer.

• Rent two more web servers, and store on each one an exact copy of your web app ready to run.

• Watch how the load balancer redirects traffic and distributes everything better during traffic spikes.

• Need more? Install more web servers, build load balancers for the load balancers, etc.

The best solution is always to hire experts in cloud web servers.

If you want to save on your cloud costs, or you’re just getting started deploying your web applications on AWS, write to me with no commitment, to my email or WhatsApp; I’ve been optimizing web applications in the cloud for all kinds of companies for more than 5 years.