Virtual Per-VM Swap Space for Live Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current virtual computing systems face inefficiencies in live migration of memory-overcommitted virtual machines due to the lack of support for per-VM swap spaces, leading to inefficient resource allocation and the need for intrusive kernel modifications.
Innovation Solution
Implementing a method that creates a virtual per-VM swap space dynamically during migration, skipping swapped pages during pre-copy migration and remapping logical addresses to new physical swap-space locations, allowing the destination host to access swapped pages on demand without kernel modifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If per-VM swap space is implemented for live migration, then migration efficiency is improved, but device complexity increases
Solution Approach 1:
The patent divides the swap space into per-VM swap spaces, where each virtual machine has its own dedicated swap space on the storage device. This segmentation allows the destination host to access only the specific swap pages needed for each VM during migration, rather than managing a monolithic swap space, thereby improving migration efficiency while keeping the complexity manageable through modular organization.
Solution Approach 2:
The patent introduces a mapping structure that acts as an intermediary between the virtual machine's memory pages and the physical swap pages on storage. This mapping layer abstracts the complexity of swap space management, allowing the hypervisor to efficiently track and migrate swapped pages without directly managing the underlying storage complexity, thus improving migration efficiency while isolating system complexity in a manageable intermediate layer.
2Reliability
If swap space is pre-allocated for each VM, then migration reliability is improved, but resource consumption increases
Solution Approach 1:
The patent implements dynamic swap space allocation where each VM's swap space is allocated on-demand based on actual swapping activity rather than pre-allocating fixed space for each VM. The system creates virtual swap files only when needed during migration, and the storage consumption adapts dynamically to the actual swap requirements of each virtual machine, ensuring migration reliability while optimizing storage resource utilization.
Solution Approach 2:
The patent changes the parameter of swap space allocation from static pre-allocation to dynamic allocation based on actual usage patterns. By monitoring which memory pages are swapped out during migration and allocating swap space only for those specific pages, the system ensures reliable migration by having the necessary swap space available while consuming only the minimum required storage resources rather than pre-reserving excessive space.
3Productivity
If kernel modifications are made to support per-VM swap space, then migration performance is improved, but ease of manufacture deteriorates
Solution Approach 1:
The patent implements a user-space swap management mechanism where the hypervisor or management software handles per-VM swap space creation and management without requiring kernel modifications. The system uses standard operating system facilities and file system operations to create and manage swap files in user space, allowing the migration infrastructure to improve performance through per-VM swap management while avoiding the complexity and security risks of kernel modifications.
Data Source
AI summary
Various embodiments disclosed herein are related to a non-transitory computer readable storage medium. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to send an indication of a first storage location to a destination host. In some embodiments, the first storage location includes content that is swapped out from a memory location in a source host. In some embodiments, the indication includes one or more of a logical address and a first physical address. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to map the logical address of the first storage location to a second physical address of a second storage location. In some embodiments, the destination host accesses the content of the first storage location.