Failure detectors were introduced as a concept to provide processes
with information about failures. We show in this paper that failure
detectors can also provide information about time. In particular, we
demonstrate that one can define a failure detector that provides
a process with the same properties as a hardware clock. This allows
us to demonstrate that all properties – except process recoveries –
of the timed asynchronous system model can be provided with the
help of . To prove this, we define a slightly restricted timed model

(which is essentially the timed model without process recov-
eries). Given an algorithm designed for , we show that the
set of runs in the timed model is the same as the set of runs
of the algorithm using the failure detector in the failure detector
model for partitionable systems.