Adaps – A three-phase adaptive prediction system for the run-time of jobs based on user behaviour

In heterogeneous and distributed environments it is necessary to create schedules for utilising resources in an efficient way. This generation often poses a problem for a scheduler, since several aspects have to be considered. One way of supporting a scheduler is to provide accurate predictions of the run-times of the submitted jobs. A large number of current techniques offer statistical models that are deployed on previously filtered data. As users have different jobs, and because the attributes of their jobs differ, filtering data and choosing an appropriate prediction method has to cover these aspects. This article describes Adaps, a system for run-time prediction that works in three phases. Each is independently adjusting to the jobs of a user, based on historical information. This leads to a user specific clustering of data and to a flexible utilisation of different prediction techniques in order to create a user-centred prediction model.