We introduce a novel hybrid failure model, which facilitates
an accurate and detailed analysis of round-based synchronous,
partially synchronous and asynchronous distributed algorithms
under both process and link failures.
Its utility and expressiveness is demonstrated by means of
a complete analysis of the well-known randomized Byzantine agreement
algorithm of (Srikanth & Toueg 1987).
Granting every process in the system up to l link
failures (with la arbitrary faulty ones among those) in every
round, without being considered faulty, we show that this algorithm
needs just n >= 4*l+ 2*la +3*fa +1$ processes for tolerating fa
Byzantine process failures. The probability of disagreement after R
iterations is only 1/2^R, which is the same as in the FLP model
and thus much smaller than the lower bound O(1/R) known for
synchronous systems with lossy links. We also provide a detailed
analysis of the algorithm's running time, as well as an evaluation 
of the model's assumption coverage in systems with transient 
link failures. Finally, stubborn links are shown to be sufficient for
this algorithm. In accordance with our findings for synchronous
systems obtained elsewhere, our results reveal that randomized
Byzantine agreement can be solved efficiently even in asynchronous
systems with imperfect communications. Contrasting widespread
believe, there is no need to employ a perfect communications
subsystem even in case of excessive link failure rates.