A platform for gathering multilayer runtime data in a datacenter

There are different kinds of datacenter monitor system, but most of them just monitor a few aspects of runtime data. Some of them just focused on temperature monitoring and power consumption. And others are focusing on application behaviors and system resources. So they can't reflect the running state of a datacenter precisely. In this paper, we introduce a platform which aims at monitoring multilayer runtime data from the underlying hardware to upper application layers. Based on these data, we evaluate the health of datacenter and provide a set of alarm mechanism.