Leveraging phase change memory to achieve efficient virtual machine execution

Abstract

Virtualization technology is being widely adopted by servers and data centers in the cloud computing era to improve resource utilization and energy efficiency. Nevertheless, the heterogeneous memory demands from multiple virtual machines (VM) make it more challenging to design efficient memory systems. Even worse, mission critical VM management activities (e.g. checkpointing) could incur significant runtime overhead due to intensive IO operations. In this paper, we propose to leverage the adaptable and non-volatile features of the emerging phase change memory (PCM) to achieve efficient virtual machine execution. Towards this end, we exploit VM-aware PCM management mechanisms, which 1) smartly tune SLC/MLC page allocation within a single VM and across different VMs and 2) keep critical checkpointing pages in PCM to reduce I/O traffic. Experimental results show that our single VM design (IntraVM) improves performance by 10% and 20% compared to pure SLC-and MLC-based systems. Further incorporating VM-aware resource management schemes (IntraVM+InterVM) increases system performance by 15%. In addition, our design saves 46% of checkpoint/restore duration and reduces 50% of overall IO penalty to the system.

Balloonfish: Utilizing morphable resistive memory in mobile virtualization

“…MMS contains a Memory Monitoring hardware that tracks memory requirement at runtime to trades off latency with capacity in an SLC/MLC PCM memory system. Meanwhile, Zhou et al proposed a cooperative cross-layer software/hardware approach, VM-aware PCMbased memory system design, in which PCM page management unit makes decisions based on the full-system execution characteristics [11]. Our work differs from the work proposed in [9,11], since our PCM page management makes decisions based on the application-specific feature and the workload memory requirement.…”

“…Meanwhile, Zhou et al proposed a cooperative cross-layer software/hardware approach, VM-aware PCMbased memory system design, in which PCM page management unit makes decisions based on the full-system execution characteristics [11]. Our work differs from the work proposed in [9,11], since our PCM page management makes decisions based on the application-specific feature and the workload memory requirement. Instead of monitoring the access information on physical addresses, our design dynamically tuning the mode of page according to the application-specific feather.…”

“…In the Linux kernel, we implement a kernel module to model the behavior and timing for PCM. Table I lists related configuration parameters, including the access latency we adopted for SLC, MLC and SD card [9][10][11].…”

“…At the software level, Dong et al [10] proposed an adaptive MLC/SLC reconfigurable PCM design for file storage. For virtualized environments, Zhou et al [11] proposed a VM-aware PCM-based system, which periodically tracks page accesses to determine the optimal SLC/MLC page ratio. These designs reap the benefits of both SLC and MLC, but they only focus on monitoring workloads to determine the optimal SLC/MLC page ratio, without considering the memory access patterns which also significantly affect the choice of using PCM as SLC or MLC: ideally, a page should consist of SLC cells when it is being actively accessed for good performance; for the rest of the time it should consist of MLC cells to save more cells as SLC for heavily accessed pages, as a page of the same size requires fewer MLC.…”

NVM heaps for accelerating browser-based applications

“…Other efforts are focused on moving a major portion of sandboxing mechanisms to the OS, similar to Android [7], but for OS agnostic applications like browsers, completely relying on OS-based isolation seems unlikely. Related work like [25] uses NVM to optimize storage in virtualized environments. Finally, a recent work on exceptionless system calls [19] studies the impact of reducing system call blocking costs and we believe such kernel techniques can be very useful in our future work.…”