Glossary · Automation software engineering and architecture
Hard real time
Also known as: Hard real-time system
German: Harte Echtzeit
In real-time systems, hard real time describes timing requirements under which every deadline must be met, because missing even one is considered a system failure that can cause damage, loss of control or hazardous behavior.
- Software engineering
In one sentence
In hard real time, every deadline must be met; missing one counts as a system failure.
Example
The current controller of a servo drive runs in hard real time: it must complete every control cycle within its fixed period, or the drive trips.
How it applies
- Engineering: Hard real-time systems need bounded worst-case execution times, deterministic scheduling, controlled interrupt latency and careful avoidance of unbounded operations, such as dynamic memory allocation or garbage collection during operation.
- Verification: Showing that deadlines are always met requires worst-case analysis or extensive measurement under worst-case load, not just average timing. Watchdogs detect overruns at runtime and trigger a defined reaction.
- Documentation: Specify hard real-time requirements with numbers and document how they were verified. Service documentation should explain cycle time monitoring and what to do after a timing fault.
Hard real time vs. firm and soft real time
Firm real time tolerates occasional misses as long as late results are discarded; soft real time tolerates late results with reduced value. Real time is about predictability, not speed: a slow system can be hard real-time if its deadlines are always met.
Timing of safety functions is part of their specification and is verified as part of functional safety work.