An execution environment consisting of virtual machines (VMs) interconnected with a virtual overlay network can use the naturally occurring traffic of an existing, unmodified application running in the VMs to measure the underlying physical network. Based on these characterizations, and characterizations of the application's own communication topology, the execution environment can optimize the execution of the application using application-independent means such as VM migration and overlay

more »

... logy changes. In this paper we demonstrate the feasibility of such free automatic network measurement by fusing the Wren passive monitoring and analysis system with Virtuoso's virtual networking system. We explain how Wren has been extended to support online analysis, and we explain how Virtuoso's adaptation algorithms have been enhanced to use Wren's physical network level information to choose VM-to-host mappings, overlay topology, and forwarding rules. 1. Monitor the application's traffic to automatically and cheaply produce a view of the application's network demands. We have developed a tool, VTTIF [2], that accomplishes this. 2. Monitor the performance of the underlying physical network by use the application's own traffic to automatically and cheaply probe it, and then use the probes to produce characterizations. This paper describes how this is done. 3. Adapt the application to the network to make it run faster or more cost-effectively. This paper extends our previous adaptation work [18, 17] with algorithms that make use of network performance information. 4. Reserve resources, when possible, to improve performance [7, 8] . Virtuoso is capable of accomplishing these feats using existing, unmodified applications running on existing, unmodified operating systems. We build on the success of our Wren passive monitoring and network characterization system [21, 20] to accomplish (2) above. Wren consists of a kernel extension and a userlevel daemon. Wren can: