Skip to main content
Institute of Advanced Technological and Commercial Studies

Resources

From prototype to production

A prototype shows an idea works; going to production commits that it will work tomorrow morning, for everyone, without you. Between the two sit precise decisions, and they can be learned.

What production adds to the prototype

In production, the service runs while you sleep. It therefore needs a record of what it does, an alert when it drifts, a restorable backup and a path back to the previous version. These four elements are designed alongside the feature, never after: an application deployed without logging is diagnosed by questioning users one at a time, which costs days where one log line costs minutes.

  • A reproducible container image, built identically every time
  • An automated deployment, repeatable and reversible
  • Structured logging, readable by a machine
  • Monitoring that alerts on drift, not on an incident already reported
  • A backup whose restore has been tested

The chain, from repository to service

  1. 01

    Build an identical image

    The container freezes dependencies. The version that passes the tests is exactly the one that ships.

  2. 02

    Deploy without manual steps

    An automated deployment replays identically. That is what makes rollback fast and safe.

  3. 03

    Observe the running service

    Structured logs, load indicators and threshold alerts: the service reports its own health.

  4. 04

    Restore and resume

    A tested backup and a rehearsed switchover turn an incident into a short, controlled interruption.

The volume the catalogue devotes to it

Dedicated block of the long computing programme.

taught hours on operations
180

Recovery plan, backups, switchover drills.

hours on service continuity
120

Hosting, continuous integration, monitoring.

hours on web deployment
140

Questions about operations

Do I need a dedicated team to operate a service?

One well-equipped person is enough for a modest service, provided the automation is in place. That is exactly what the operations block installs: repetitive gestures become scripts, and human attention concentrates on diagnosis.

What is the difference between logging and monitoring?

The log tells what happened, line by line, and serves diagnosis. Monitoring aggregates indicators and raises an alert on a threshold. The two complement each other: the alert calls you, the log explains.

When should a prototype move to production?

When someone is about to depend on it for their work. At that precise point the four operational elements become necessary, and adding them before the switch costs far less than grafting them onto a service already in use.

From prototype to production — Resources | IHETC — IHETC