Technology

From the edge to the record.

Astro Interactive runs the platforms it owns. A request crosses a public boundary, reaches a service that knows what to do with it, and, when it must, changes something we are willing to keep. This page is that path. It is not a map of private hosts.

  1. 01 Edge What is allowed in.
  2. 02 Services What the product does.
  3. 03 Data What has to last.
  4. 04 Operations What keeps it honest.

01

Edge

The public side of a platform is a boundary, not a hallway. Only the paths that should be reachable are reachable. Gateways and firewall policy are written as part of the design, and DNS follows the domains the product actually uses.

Traffic is watched where it matters: enough to notice abuse and failure, not so much that the network becomes a second product. Application checks sit behind that boundary. Secrets stay out of the page, the repository, and this description. The internal layout is not published, on purpose.

02

Services

Behind the edge, work is split where the work actually splits. Authentication, product APIs, background jobs, and shared utilities are separate services with contracts between them. One platform can change without quietly taking another with it.

Those contracts are how Astro Interactive products talk to clients and to each other. They are authenticated where they need to be, and a failure says what failed. An operator should be able to read the response and know the next step, rather than guessing which box went quiet.

03

Data

The runtime follows the workload. A long-running game process, a web origin, and a worker at the edge are not the same machine wearing different names. Each is kept as a production system, and its configuration is part of that system rather than a note someone remembers.

Persistence is chosen for the way the data is used. Access belongs to the service that needs it. Moving a schema, and recovering one, is decided before an incident makes the decision for us. Backups are an operational requirement, not a feature we mention after the fact.

04

Operations

A release is a path: build, publish, verify. Environments stay comparable, so a rollback is something we can actually do. Automation takes the repetitive work, and the dangerous work, off of memory.

Health checks, status pages, logs, and alerts exist so a failure shows up for the people who run the platform before it shows up as a complaint. Monitoring is part of shipping. It is not a dashboard added once the launch post is already written.

Ask about capability