puters would behave as programmed. Everyone knows this to be simplistic; yet still, a system administrator would worry about behavior only when an incident reported something to the contrary.

IllustratIon by stuart braDForD

from Demolition to Maintenance
When a problem occurs, many orga-
nizations take affected systems out
of service, wipe them, and restore
them from backup or reinstall from
scratch. This is the only way they
know to assure the state of the system
because they know no simple way of
discovering what changed without an
arduous manual investigation. The
process is crude, like tearing down a
building to change a lightbulb. But
the reason is understandable. Cur-
rent tools are geared for building, not
repairing—and if you’ve only got a
sledgehammer...then you rebuild.