Academic Journal
DTM: Memory state-aware dynamic tiering for cloud virtual machine with CXL-enabled tiered memory.
| Τίτλος: | DTM: Memory state-aware dynamic tiering for cloud virtual machine with CXL-enabled tiered memory. |
|---|---|
| Συγγραφείς: | Liu, Rongkai1 (AUTHOR) liurk1@chinatelecom.cn, Peng, Jing2 (AUTHOR) pengjing.bj@chinatelecom.cn, Gao, Xiang1 (AUTHOR) gaox15@chinatelecom.cn, Lin, Xiancheng1 (AUTHOR) linxc4@chinatelecom.cn, Zhu, Yuanrui1 (AUTHOR) zhuyr1@chinatelecom.cn, Wu, Yuting1 (AUTHOR) wuyt6@chinatelecom.cn, Lu, Gang1 (AUTHOR) lugang@chinatelecom.cn |
| Πηγή: | Future Generation Computer Systems. Oct2026, Vol. 183, pN.PAG-N.PAG. 1p. |
| Θεματικοί όροι: | *Virtual machine systems, *Mathematical optimization, Computer memory management |
| Περίληψη: | CXL-based disaggregated memory, which enables on-demand allocation and flexible integration with high-performance local memory, offers significant potential for cloud computing. However, workload heterogeneity and the complexity of the cloud software–hardware stack introduce substantial performance challenges for tiered memory systems. Existing studies primarily focus on tiering management within the host operating system or hypervisor, leaving cloud users unaware of memory access patterns and scheduling decisions. This black-box design prevents users from optimizing tiered memory configurations according to workload characteristics. To address this limitation, we propose a memory-state analytical model that captures page flow dynamics and access distributions across memory tiers, characterizing the relationship between optimization objectives and tiering strategies. Leveraging this model, we design a memory-state-aware tiering algorithm that dynamically adapts page migration decisions to workload characteristics. Based on the proposed model and algorithm, we develop DTM, a user-level dynamic tiering framework for virtual machines equipped with tiered memory. DTM collects memory-state information in user space, invokes the tiering algorithm, and dynamically controls page migration according to optimization objectives. We evaluate DTM using representative cloud workloads under diverse tiered-memory configurations. For single-VM scenarios, DTM consistently improves performance for memory-bound workloads, achieving average performance gains ranging from 8.0% to 20.2% compared to state-of-the-art tiering systems. In concurrent multi-VM environments that emulate realistic cloud deployments, DTM also delivers substantial performance improvements. [ABSTRACT FROM AUTHOR] |
| Copyright of Future Generation Computer Systems is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Βάση Δεδομένων: | Business Source Index |
καταχωρήστε σχόλιο πρώτοι!