Holistic Performance Monitoring of Hybrid Clouds: Complexities and Future Directions

In a cloud environment, computing resources and applications are provided as services over a network, typically the Internet. Cloud computing resources are largely hosted in commercial datacenters and colocation facilities and in hyperscale server farms operated by companies like Amazon, Apple, Google, Microsoft, and Facebook. These datacenters must be continuously monitored to ensure stable operation. Various monitoring tools are available to monitor different layers of an enterprise IT system-from business functions to applications to infrastructure. However, determining what, when, and where to monitor is left to the system administrators. The intuition-driven monitoring strategy is therefore ad hoc, of variable quality, and not scalable to large systems. Furthermore, this approach fails to keep up with the continuous evolution of enterprise systems, especially in hybrid cloud computing environments, which integrate multiple public and private datacenters.