Unstoppable Hardware
Why Leaving ENIAC Turned On Forever Made It More Reliable
The ENIAC was a machine built out of nearly 18,000 vacuum tubes, which meant it was also a machine perpetually on the verge of collapsing under its own heat. Those delicate glass-enclosed filaments burned out with astonishing regularity, keeping engineers in a state of perpetual emergency. Every burnout threatened to plunge the colossal machine back into darkness.
Yet the most dangerous moment for a vacuum tube wasn't when it was working at full capacity. It was the moment the machine was switched off.
When early electronic computers powered down, the sudden drop in temperature sent a shockwave of contraction through the hardware. When they powered back up, the reverse happened. This thermal shock violently stressed the delicate metal components inside the tubes, causing brittle welds and filaments to snap under the rapid expansion. The machine was essentially tearing itself apart simply by being turned on and off.
Engineers led by John Presper Eckert and John Mauchly found a remarkably simple workaround: they stopped turning it off. By keeping the ENIAC powered on continuously, they smoothed away the violent temperature swings that caused most of its failures. The constant heat proved far kinder to the hardware than the cycle of rest.
This continuous-power policy quietly transformed the nature of computing. Maintenance shifted away from frantic, round-the-clock crisis management into predictable, scheduled replacements, offering an early blueprint for how we think about high-availability systems today.
Reliability in complex hardware is rarely about stopping things from aging. Instead, it is about surviving the transitions.
Published