Answers are grounded in this paper and cite their sources.
Demonstrates improved memory efficiency in virtual machines, suggesting enhanced cloud competitiveness.
In virtualization-based cloud computing, numerous redundant tiered memory pages exist across independent virtual machines (VMs) deployed on different physical hosts within the data center. For instance, each VM typically loads the data from its virtual disk into its memory (local tier), without considering the existing data in other remote VMs’ memory (remote tier). In this paper, we propose FlexMem, a preliminary architecture designed to eliminate memory redundancy for qcow2-based VMs. First, FlexMem introduces a virtual persistent memory (PMem) device that facilitates cross-host memory consolidation by leveraging Remote Direct Memory Access (RDMA) for fast remote read operations. Second, FlexMem employs flexible backing specifications that are memory-mapped (mmapped) to sparse and layered qcow2 images. By delivering images to VMs through FlexMem and enabling guest Direct Access (DAX), user programs within the guest can directly access identical images already cached in the remote host’s memory. Our evaluation demonstrates that FlexMem significantly improves memory efficiency and the availability of VMs with reduced hardware costs, ultimately enhancing the competitiveness of cloud vendors.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: