Seshat: uma arquitetura de monitoração escalável para ambientes em nuvem.

Abstract

System monitoring has an essential role in management, maintenance and to provide feedback through data collection. Particularly, cloud monitoring must address several virtualized resources of dynamic behavior, such as infrastructure performance metrics, application logs, and data produced in distributed environments. To meet all of these requirements and deal with complexity involved in cloud systems is certainly a challenge. This work has investigated the needs of a cloud monitoring system and after identifying some essential roles, we propose a general use architecture to supply them. The layered cloud monitoring architecture is elastic, scalable, resilient and extensible. Lastly, we apply knowledge obtained designing a monitoring system based on open source tools that deploys the proposed architecture. As a monitoring subject, we have selected a virtualized big data environment that is usually deployed in clouds. Our experiments proved the system's scalability, achieving up to half the delay time in message reception. Additionally, we have included a number of use cases related to our experiences with the system.