Packet probing as network load detection for scientific applications at run-time

Summary form only given. High-performance applications place great demands on the computation and communication resources of a distributed computing platform. If the availability of the resources changes dynamically, the application performance may suffer. This is especially true for cluster environments, which are often heterogeneous and require tedious tuning for high-performance applications. We describe a packet probing technique to detect contention on the cluster nodes to which application is mapped. The technique is light-weight and may be a priori tuned to a given network type, so that it is used at an application's run-time. We also show an easy integration of packet probing as a module of a recently developed communication middleware, which provides an application with a run-time access to the dynamic computing system information and which invokes application adaptations.