Virtual CPU state detection and execution flow analysis by host tracing

Cloud computing offers to the end user the abilityof accessing a pool of resources with the Pay as Use (PaU)model. By leveraging this technology, users can benefit fromhardware virtualization for on-demand resource acquisition andrapid elasticity. However, there is no effective tool to analyzevirtual hardware performance, especially when isolation betweenthese virtual resources is not adequate. The existing tools need toaccess and trace the whole activity of the VM and host. However,in most cases, tracing the virtual machine (VM) is not possiblebecause of security issues and the added overhead. Therefore,there is a need for a tool to troubleshoot unexpected behavior ofVMs without internal access for tracing or debugging.In this paper, we propose a new method to study the state ofCPUs inside VMs without internal access. Our tool can detectunexpected delays and their root causes. We developed a virtualCPU (vCPU) state analyser to detect the state of vCPUs alongwith the reason for being in that state. This approach relies onhost tracing, thus adding less overhead to VMs as comparedto existing approaches. Then we propose a new approach forprofiling threads inside the VMs by host tracing. We implementeddifferent views for the TraceCompass trace viewer to let theadministrator visually track different threads and their statesinside the VMs. Our tool can detect different problems such asovercommitment of resources.

Uncontrolled Keywords

virtualization; KVM; performance analysis; vCPU states; LTTng; TraceCompass