PTEC Storage

PTEC Storage is the file-delivery layer for library.ptec.edu.kh. Its landing page states that and nothing else: "File delivery service for library.ptec.edu.kh. There is nothing to see here." No catalogue, no navigation, nothing to browse — and a noindex directive so it stays out of search results.

That is the whole design. This is infrastructure, and the most useful thing infrastructure can do is stay out of the way.

Directly, the library application. Indirectly, every reader of the PTEC Digital Library — none of whom ever sees this service by name.

One job, done narrowly: deliver the library's files over a separate origin, with caching and hardening tuned for static assets rather than for an application.

I work on this service as part of the PTEC Digital Library platform, on the infrastructure side. Its internal design — storage backend, topology and operational tooling — is intentionally not published here.

The service has no user interface to protect and no reason to be crawled. That is a constraint worth having: it allows the origin to be locked down far harder than an application ever could be.

There is exactly one interface decision, and it is to have almost no interface. The landing page names the service, names the site it serves, and says there is nothing else here — which is more useful to someone who lands on it by accident than either a redirect or a 404 would be.

A separate origin from the library application, served through Cloudflare. The library treats it as two things at once and its Content-Security-Policy says so: storage-ptec.online and its subdomains are permitted image sources, and api.storage-ptec.online is a permitted connect-src origin. Splitting delivery from rendering lets the two be cached, scaled and hardened on their own terms.

This service is defined more by what it refuses than by what it offers.

The policy is deny-by-default and the strictest of the three platforms: default-src 'none', img-src 'self', frame-ancestors 'none', base-uri 'none', form-action 'none' — no scripts, no framing, no forms, no base-tag rewriting. Alongside it: HSTS, X-Content-Type-Options nosniff, X-Frame-Options DENY, and a Permissions-Policy that switches off camera, microphone, geolocation, payment and USB. Because there is no UI to protect, everything that is not file delivery is simply turned off. Network error reporting is configured, so failures are visible rather than silent.

The landing page is one heading and one sentence in a system font stack at default sizes — legible without CSS and trivial for a screen reader to read out. There is nothing else on it to make accessible.

Deliberately invisible: the landing page is marked noindex. A file host has no business competing with the library it serves for that library's own search results.

Responses are cached publicly for five minutes and carry Last-Modified, so a repeat request revalidates cheaply instead of re-downloading. Accept-Ranges: bytes is advertised, which is what lets a reader resume an interrupted download or a viewer seek into a large PDF rather than fetching all of it first. Cloudflare fronts the origin, and HTTP/3 is advertised.

Separating file delivery from an application is easy to describe and easy to get wrong. The split only pays off if the second origin is genuinely independent — its own caching, its own policy, its own failure mode — rather than a proxy that inherits the first one's constraints.