在現代云計算平台中，高可用性（HA）是重要的方面之一，該平台應確保在其計算池上運行的服務的活動性(liveness)，並在服務故障時自動進行故障轉移。現代的雲計算平台通常使用系統層心跳(system layer heartbeating)來檢測平台中的故障。先前的研究（在OpenStack上實施的高可用性軟件系統（HASS））應用了多個檢測器，並被證明比系統層心跳更有效率。但是，HASS在另一種可能具有不同支持的檢測器的環境(異質性環境，heterogeneous environment)中不太可靠。異質性環境中的異構計算主機將影響受支持的檢測器。而且在某些情況下，檢測器可能無法正常工作，或者在某些環境中可能會丟失。在上述情況下，需要人為乾預以更改原始故障模型或修復檢測器故障。因此，本研究旨在提供一種機制，該機制自動地重新配置故障模型以為雲計算平台繼續提供活動性檢測。也就是說，即使某些檢測器出現故障或丟失還是能繼續提供HA保護。使用實驗方法，有兩個主要結果：（i）所提出的機制與系統層心跳一樣可靠； （ii）雖然在最壞的情況下檢測時間與系統層心跳相等，但其他情況下，所提出的機制比系統層心跳更有效率。;In a modern cloud computing platform, High availability (HA) is one of important aspects where the platform should ensure the liveness of services that runs on its compute pool and do failover automatically when services are off. A modern cloud computing platform commonly uses system layer heartbeat to detect faults in the platform. The previous research, High-Availability Software System (HASS) implemented on OpenStack, applied several detectors and proven to be more efficient than heartbeat method. However, HASS not so reliable in another environment which may has different supported detector. Heterogenous computing host that built the environment will affect the supported detector. In some conditions, detector may be failed to works properly, or it may not available in some environments. In this case, human intervention is needed, either to change the original fault model or to fix the detector fault or use the common method to detect all kind of faults. Therefore, this research aims to provide an efficient mechanism that applicable in heterogenous cloud computing platform to continuously provide liveness detection even if some detectors are failed or not available. Using experimental method, there are two main results, which are: (i) the proposed mechanism is as reliable as heartbeat method; and (ii) the proposed mechanism is more efficient than heartbeat method, although in the worst case the detection time is near equal to heartbeat method.
