Runtime Measurements

The measurements are made by setting a time stamp t_before before the method calls of adaptor methods and afterwards t_after. The difference between t_before and t_after is taken for the runtime of the method. For the time stamps the global time is taken. In order to get reproducable results, the measurements are done on a single-user host, where the conditions for the measurements can be hold stable for the period of measuring and restored as needed.

Time stamps within the adaptation mechanisms measure the time needed for context awareness and remote class loading for a more detailed analysis of the method runtime. The implementation of the customized version contains only time stamps for the method calls of the adaptor classes. The time which is needed for migration is not measured because the effort for setting up synchronized clocks between the hosts is much higher than the gain of exact results compared to a more simple approximation. The lowest time which is needed for migration is the size of executable code divided by the theoretical bandwidth.