Virtual Machine Migration With Bitmap-Based Memory Placement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current virtual machine migration methods fail to distinguish between active and inactive data during memory migration, leading to inefficient use of volatile memory at the destination host and increased data access latency.
Innovation Solution
A method and device that utilize a memory bitmap to segregate data into volatile and non-volatile memories based on their usage patterns, ensuring that active data is stored in volatile memory and inactive data in non-volatile memory, maintaining consistent distribution and reducing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the source host directly packs and sends all memory data to the destination host without distinguishing between active and inactive data, then the migration process is simple, but the volatile memory space at the destination host is wasted and active data may be stored in non-volatile memory causing increased access latency
Solution Approach 1:
The patent segments memory data into two categories: active data (frequently accessed) and inactive data (less frequently accessed). This segmentation is achieved by dividing the memory into multiple memory units and using a memory bitmap to identify which units contain active data. During migration, active and inactive data are handled differently to optimize both migration efficiency and post-migration performance.
Solution Approach 2:
The patent performs preliminary identification of active and inactive data units before the actual migration process. The memory bitmap is generated and transmitted to the destination host in advance, allowing the destination host to prepare appropriate storage locations (volatile memory for active data, non-volatile memory for inactive data) before the data arrives, thus optimizing access latency from the start.
2Speed
If the destination host stores received memory data into volatile memory, then data access speed is improved, but idle storage space in volatile memory decreases gradually and may need to overflow to non-volatile memory
Solution Approach 1:
The destination host receives the memory bitmap before or during the migration process and uses it to pre-allocate appropriate storage locations. Inactive data units are directed to non-volatile memory while active data units are directed to volatile memory, preventing volatile memory from filling up and avoiding the need to store active data in slower non-volatile memory.
Solution Approach 2:
The patent applies different storage quality characteristics to different parts of the memory data based on their access patterns. Active data units are stored in high-speed volatile memory with fast access characteristics, while inactive data units are stored in non-volatile memory. This local differentiation of storage quality optimizes overall system performance by ensuring frequently accessed data resides in faster memory.
3Quantity of substance
If active data is stored in non-volatile memory at the destination host, then storage capacity is sufficient, but the virtual machine experiences increased data access latency affecting running speed
Solution Approach 1:
The patent applies different storage media qualities to different data units based on their access characteristics. By using the memory bitmap to identify active data units, the system ensures these are stored in volatile memory with fast access speeds, while inactive data units are stored in non-volatile memory. This local quality differentiation maintains virtual machine running speed by keeping hot data in fast memory while still providing sufficient total storage capacity.
Solution Approach 2:
The memory space is segmented into portions for volatile memory and non-volatile memory based on the memory bitmap analysis. This segmentation allows the system to allocate storage resources efficiently, placing active data in volatile memory to maintain running speed while using non-volatile memory for inactive data to ensure sufficient total storage capacity.
Data Source
AI summary
A first device receives a migration instruction sent by a second device and creates a target virtual machine of a to-be-migrated virtual machine that is in the first device. Then, the first device receives a memory bitmap of the to-be-migrated virtual machine from the second device, where the memory bitmap may indicate whether data of each memory unit in a memory of the to-be-migrated virtual machine is stored in a non-volatile memory or a volatile memory. In a virtual machine migration process, the first device stores, based on the memory bitmap, data in the memory of the to-be-migrated virtual machine into a volatile memory and a non-volatile memory of the target virtual machine.