Cloud computing offers many benefits, but IT staff lose the ability to “see” behind the cloud, jeopardizing their ability to troubleshoot problems and assure performance meets application SLAs. What happens when an application slows? In large networks with physical addresses for everything, troubleshooting performance problems is often a nightmare. Moving everything into the cloud just makes it more difficult. Fortunately, there are best practices that allow the cloud infrastructure to be instrumented, monitored, and constantly, yet unobtrusively, measured to see how it is affecting applications. At the same time, these practices produce the metrics necessary for granular SLAs.