A key benefit of Amazon EC2-style cloud computing service is the ability to instantiate a large number of virtual machines (VMs) on the fly during flash crowd events. Most existing research focuses on the policy decision such as when and where to start a VM for an application. In this paper, we study a different problem: how can the VMs and the applications inside be brought up as quickly as possible? This problem has not been solved satisfactorily in existing cloud services. We develop a fast start technique for cloud applications by restoring previously created VM snapshots of fully initialized application. We propose a set of optimizations, including working set estimation, demand prediction, and free page avoidance, that allow an application to start running with only partially loaded memory, yet without noticeable performance penalty during its subsequent execution. We implement our system, called Twinkle, in the Xen hypervisor and employ the two-dimensional page walks supported by the latest virtualization technology. We use the RUBiS and TPC-W benchmarks to evaluate its performance under flash crowd and failure over scenarios. The results indicate that Twinkle can provision VMs and restore the QoS significantly faster than the current approaches.

Hypervisor-based fault tolerance (HBFT), which synchronizes the state between the primary VM and the backup VM at a high frequency of tens to hundreds of milliseconds, is an emerging approach to sustaining mission-critical applications. Based on virtualization technology, HBFT provides an economic and transparent fault tolerant solution. However, the advantages currently come at the cost of substantial performance overhead during failure-free, especially for memory intensive applications. This paper presents an in-depth examination of HBFT and options to improve its performance. Based on the behavior of memory accesses among checkpointing epochs, we introduce two optimizations, read-fault reduction and write-fault prediction, for the memory tracking mechanism. These two optimizations improve the performance by 31 percent and 21 percent, respectively, for some applications. Then, we present software superpage which efficiently maps large memory regions between virtual machines (VM). Our optimization improves the performance of HBFT by a factor of 1.4 to 2.2 and achieves about 60 percent of that of the native VM.

Hypervisor-based fault tolerance (HBFT), a checkpoint-recovery mechanism, is an emerging approach to sustaining mission-critical applications. Based on virtualization technology, HBFT provides an economic and transparent solution. However, the advantages currently come at the cost of substantial overhead during failure-free, especially for memory intensive applications. This paper presents an in-depth examination of HBFT and options to improve its performance. Based on the behavior of memory accesses among checkpointing epochs, we introduce two optimizations, read fault reduction and write fault prediction, for the memory tracking mechanism. These two optimizations improve the mechanism by 31.1% and 21.4% respectively for some application. Then, we present software-superpage which efficiently maps large memory regions between virtual machines (VM). By the above optimizations, HBFT is improved by a factor of 1.4 to 2.2 and it achieves a performance which is about 60% of that of the native VM.