Opvis: extensible, cross-platform operational visibility and analytics for cloud

Operational visibility is an important administrative capability and a critical factor in deciding the success or failure of a cloud service. It is becoming increasingly complex due to the need to (1) track both persistent and volatile system state across heterogeneous endpoints and (2) consider a broader range of data sources fueled by demand for sophisticated analytics. In this paper we present OpVis, a monitoring and analytics framework to provide operational visibility without the limitations of traditional fragmented monitoring solutions. We highlight OpVis' extensible data model, enabling custom data collection and analytics based on the cloud user's requirements, describe its monitoring and analytics capabilities, present performance measurements, and discuss our experiences while supporting operational visibility in our cloud.