Method and system for exchanging local and remote memory objects in single copy mode

By implementing a memory object swapping method and system in single-copy mode, the problems of data redundancy and address switching overhead in remote memory systems are solved, achieving efficient memory resource utilization and performance scheduling, and meeting the needs of modern data centers.

CN122363877APending Publication Date: 2026-07-10ZHEJIANG UNIV +1
0 Cites 0 Cited by

Patent Information

Application Number
CN202610273804.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-03-07
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing remote memory systems often employ a multi-copy model, leading to data redundancy, low memory utilization, and bottlenecks such as large address space switching overhead and difficulty in handling high-frequency requests when multiple processes share memory, severely limiting system scalability and operating efficiency.

Method used

The memory object swapping method and system adopts a single-copy mode. By unifying the remote memory address space, using a unique copy mode and a dedicated memory allocator, it achieves physical separation between the metadata describing the memory block and the actual data of the object. It offloads memory allocation operations to a dedicated CPU core and uses a multi-partition load balancing strategy to dynamically schedule memory requests.

Benefits of technology

It achieves efficient utilization of remote memory resources, improves performance scheduling in multi-process sharing scenarios, adapts to the application needs of modern data centers, avoids inter-process address switching overhead and data redundancy, and improves memory allocation efficiency.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a method and a system for exchanging local and remote memory objects in a single copy mode. The method comprises the following steps: firstly, constructing a logic address space, deploying the logic address space on a memory node, establishing a corresponding relationship between a unified space address and a virtual memory address, and realizing conversion from the virtual memory address to a physical address through a memory management unit; secondly, triggering object swap-out and swap-in operation by adopting a unique copy mode, and balancing memory release efficiency and operation stability through an asynchronous recovery mechanism; and finally, the memory management efficiency is improved through a special memory allocator, including separated memory layout, a core unloading mechanism and multi-partition load balancing. According to the method, the inter-process address switching overhead is avoided by unifying the address space; eliminating data redundancy and multi-copy synchronization overhead by adopting a unique copy mode; the memory allocation efficiency is improved by utilizing separated memory layout, core unloading and a multi-partition management mechanism, and efficient utilization of remote memory resources and high-performance scheduling in a multi-process sharing scene are realized.
Need to check novelty before this filing date? Find Prior Art