High availability resource monitoring solution using simplified team formation algorithm

摘要

One of the greatest challenges that Resource Monitoring System (RMS) needs to overcome is to have higher availability compare to the system that it monitors. This is because RMS oversees the system's performance from all aspects and raises alerts when the system is under critical situation. Conventionally, the RMS is hosted by dedicated and redundant servers, which requires human intervention to perform failback operation. This work introduces the concept of virtualized resource monitoring, and proposes a novel approach to improve RMS availability through automated service placement, automated failover and failback mechanism. A prototype of the virtualized RMS has been implemented and tested on an ICT teaching laboratory with 40 workstations environment. The result shows that the performance of workstations are not affected much by hosting the virtual RMS, yet the availability of RMS features can be maintained at a high level. From the properties and behaviors of the prototype, this paper also highlights the feasibility and future improvement of such a proposed approach.