Why has "an administrator checks the systems" stopped being a workable way to run IT infrastructure?
Because it is manual and reactive: it cannot scale to virtualised, hybrid, containerised estates, and it only discovers a fault after the fault has already hit users.
Two things happened at once. The estate grew more complex — virtualisation, hybrid and cloud infrastructure, containers, microservices — so the number of moving parts a person would have to look at exploded. And the demands on it grew — availability, performance, security and compliance are now contractual, not aspirational.
| Classic approach | Modern approach | |
|---|---|---|
| Who looks | An admin, by hand | Continuous, automated collection of every state |
| When | After something breaks | Before it breaks — anomalies spotted in advance |
| Posture | Reactive | Proactive |
| Scaling | Breaks down as complexity grows | Scales with the estate |
| Outcome | Fault fixed after the outage | Automation steers the infrastructure itself |
The payoff is not only "fewer outages". Continuous observation is what makes capacity planning (you can see the trend), auditability (you can prove what happened), SLA evidence and forensics possible at all — none of which a person spot-checking servers can deliver.
And note the direction this points: once every state is captured automatically, acting on those states automatically is the obvious next step. Observation is the precondition for automation, which is why the two belong in the same course.
Go deeper:
Google SRE Book — Introduction — why Google replaced sysadmin-style operations with engineering and automation, the same shift this card describes.