My choice of cloud

2026-02-07

Recently I have been reviewing my chosen technologies used for deploying pet projects and have found myself steering sharply away from the big dogs on the block (AWS, GCP, Azure).

My AWS Journey

My journey into the cloud started when I was a datacenter operations engineer for AWS, where my working life involved a lot of DIMM swapping and cable plugging. Seeing the cloud from the inside sparked a lot of interest, and being an employee of AWS meant that I had a sandbox of services at my fingertips free of charge (so long as no emails came to me from management).

I began the usual studying of certs but quickly saw the value and started using AWS for random projects, a lot of which were just running Python scripts in Lambda.

Shortly after, I grew my career in cloud and found myself becoming an AWS fanboy, developing bias towards my chosen cloud both as a result of having worked there and it also being the only cloud I was familiar with.

Little problems with big clouds

Though I back up my preference by genuinely viewing AWS as having the best fundamental offerings and good prices, I am aware that AWS's competitors have fine-tuned their own experiences (GCP has great K8s and a good dev workflow with projects). With each cloud provider having their own unique pros and cons, I find they all struggle with the same issues, and that is offering clean, complete services.

AWS has definitely tried with the likes of Lightsail, but my experience still always feels bogged down as a result of it all being tied under the greater AWS or GCP platform.

When using the fundamental services (i.e. Route53), I find it cumbersome to log into AWS, get overwhelmed with services (molehill complaints, I know), and head through a few other button clicks to manage my DNS. The same for my compute as well (I am an advocate for ECS and alternatives to K8s for a lot of use cases).

Finding Simpler Alternatives

This slight annoyance led me to explore options, quickly discovering the well-known Cloudflare and slightly lesser-known fly.io. This switch simplified my cloud workflow in an instant, and I haven't gone back.

Cloudflare provides me everything from DNS, CDN, object storage, compute, databases, and fly.io offers such a nice-to-use workflow for compute deployments, especially as someone who enjoys the GOTH stack (Go, Templ, HTMX). Both of these providers do so with straightforward and concise dashboards. Admittedly, the result of them not having the same amount of services available as major cloud providers, but even the big clouds know this can be a curse with them shutting down significant services each year as they create new ones.

My Views on Big clouds

My views on major cloud providers still remain the same and I still fanboy over AWS. However, for me they are reserved for the large and enterprise, as a lot of the features just aren't necessary for the projects I undertake. Even then, I somewhat only view them as providers of bulk compute and streamlined networking as opposed to platforms for particular services.

This does seem like the obvious "use the right tool for the job," but historically I felt I had to use the tools used in my day job for my startup projects and that if I didn't, I would somehow lose skills in those areas. This, of course, is not to be true and being more selective with my choices allowed me to be more productive in the projects I take on personally.

Cloud Stack

So with that said, what does my cloud stack look like now?

Fly.io

Cloudflare

MongoDB Atlas

Redis cloud

Honorable Mentions

supabase