Most Small Sites Do Not Need Edge Computing
Edge computing can cut latency, but most small business sites need a CDN and clear architecture first. Here is how to decide.
"Edge" has become a word you see on every hosting pricing page. Code runs on geographically distributed servers, close to the user, instead of a single central datacenter. The question we often get from clients: "do we need this?"
When it actually matters
For applications with global users and frequent server interactions - authentication, APIs called on every click, real-time personalization - the latency difference between edge and a single datacenter is noticeable, especially for users far from the main region.
When it's an unnecessary complication
For a landing page or a presentation site - most of our projects - the content is static or nearly static. A classic CDN in front of a static build solves 95% of the latency problem, without the added complexity of running logic at the edge (runtime limitations, harder debugging, edge-specific cold starts).
What we actually do
- Landing pages and presentation sites: static build, served through a CDN. Simple, fast, cheap.
- Applications with server logic (dashboards, forms with server-side validation): classic server, with a region chosen close to the main user base.
- Edge functions: only where we measure an actual latency problem, not preemptively.
The rule we always apply: don't add infrastructure for a problem you haven't measured. Edge computing solves a real problem for certain applications - but for many sites, it's a solution looking for a problem.
At Legacies, this is how we scope the infrastructure behind a client website or custom tool. We start with the people using it, the data it needs to move, and the actual bottleneck. A useful application should feel fast because its architecture matches the job, not because it carries an expensive label.