Cloud computing in the context of service choreographies brings new opportunities while performing the service provisioning, i.e. the capability of acquiring and releasing resources on-demand. In fact, the Cloud providers offer the not trivial task of migrating services, as well as allocating, and deallocating resources from their offered infrastructure. Currently these activities are pursued by generic load-balancing strategies that are agnostic with respect to the overall goals of a service choreography. In this sense, service choreography offers means to predict events, and to keep of the interactions among choreographed participants, so that each software services can meet its Service Level Objectives (SLOs). In literature, performance antipatterns are well-known bad design practices that lead to software systems suffering by poor performance. Revealing a performance antipattern is an indicator that an undesired configuration could be met. In this paper we discuss how performance antipatterns can be used for mastering service choreographies that are enacting on a Cloud. Specifically, the status of the underlying Cloud infrastructure is monitored in order to reveal performance antipatterns, whose resolution is combined with look-ahead information extracted from the service choreography into meaningful mitigation guidelines that can be adopted at the application level.