Analytic models of queuing networks have been observed to give good results for systems which do not conform to the assumptions of classical queuing theory. This has been explained to some extent by the recently proposed concept of operational analysis. However, the use of operational analysis for prediction of system performance involves a restrictive assumption, that of homogeneous service times (HST), and practical systems often deviate from HST behavior. It is shown that the main system performance measures are surprisingly insensitive to violations of the HST assumption. This further explains the robustness of analytical models for predicting performance of queuing networks. Some issues regarding the operational versus stochastic approaches are resolved, since it is claimed that the present analysis, which is carried out in the operational framework, cannot be meaningfully carried out in the stochastic framework.