Adaptive Real-time Monitoring for Large-scale Networked Systems

Large-scale networked systems, such as the Internet and server clusters, are omnipresent today. They increasingly deliver services that are critical to both businesses and the society at large, and therefore their continuous and correct operation must be guaranteed. Achieving this requires the realization of adaptive management systems, which continuously reconfigure such large-scale dynamic systems, in order to maintain their state near a desired operating point, despite changes in the networking conditions.The focus of this thesis is continuous real-time monitoring, which is essential for the realization of adaptive management systems in large-scale dynamic environments. Real-time monitoring provides the necessary input to the decision-making process of network management, enabling management systems to perform self-configuration and self-healing tasks.We have developed, implemented, and evaluated a design for real-time continuous monitoring of global metrics with performance objectives, such as monitoring overhead and estimation accuracy. Global metrics describe the state of the system as a whole, in contrast to local metrics, such as device counters or local protocol states, which capture the state of a local entity. Global metrics are computed from local metrics using aggregation functions, such as SUM, AVERAGE and MAX.Our approach is based on in-network aggregation, where global metrics are incrementally computed using spanning trees. Performance objectives are achieved through filtering updates to local metrics that are sent along that tree. A key part in the design is a model for the distributed monitoring process that relates performance metrics to parameters that tune the behavior of a monitoring protocol. The model allows us to describe the behavior of individual nodes in the spanning tree in their steady state. The model has been instrumental in designing a monitoring protocol that is controllable and achieves given performance objectives.We have evaluated our protocol, called A-GAP, experimentally, through simulation and testbed implementation. It has proved to be effective in meeting performance objectives, efficient, adaptive to changes in the networking conditions, controllable along different performance dimensions, and scalable. We have implemented a prototype on a testbed of commercial routers. The testbed measurements are consistent with simulation studies we performed for different topologies and network sizes. This proves the feasibility of the design, and, more generally, the feasibility of effective and efficient real-time monitoring in large network environments.QC 2010072