Cross-Host Memory Sharing via Address Translation for VM Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face challenges in optimizing hardware resource allocation due to varying task requirements, leading to inefficient utilization rates as processing units or memory spaces remain idle due to mismatched resource demands.
Innovation Solution
A computing system and method enabling memory sharing among hosts through memory extension devices that facilitate address translation, allowing flexible configuration of hardware resources by accessing non-consecutive and fragmented memory spaces across multiple hosts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a cloud computing system configures hardware resources strictly according to task requirements, then each task can be processed correctly, but hardware utilization rate deteriorates due to idle processing units or memory spaces
Solution Approach 1:
The patent merges memory resources from multiple hosts into a shared memory pool that can be dynamically allocated across different virtual machines and tasks. Instead of dedicating memory to single hosts, the system combines available memory spaces from multiple physical hosts and allows flexible allocation, ensuring that processing units can always find matching memory resources even when individual hosts have idle memory, thereby resolving the contradiction between correct task processing and hardware utilization
Solution Approach 2:
The patent implements dynamic memory allocation where memory resources are not statically assigned but can be dynamically adjusted based on real-time task requirements and host availability. The system can dynamically allocate memory from any available host to any virtual machine needing resources, and can dynamically reconfigure allocations as tasks complete or new tasks arrive, maximizing hardware utilization while ensuring tasks receive appropriate resources
2Quantity of substance
If a task requires larger memory space, then memory capacity is improved, but processing units remain idle due to lack of configurable memory space
Solution Approach 1:
The patent makes memory resources universal by allowing any host's memory to serve any virtual machine's needs. Instead of memory being tied to a specific host, the system creates a universal memory pool where memory spaces from multiple hosts can be allocated to any processing unit across the system. This multi-functionality ensures that processing units can always access required memory capacity regardless of which physical host the memory originates from, eliminating idle processing units
3Power
If a task requires larger number of processing units, then processing capacity is improved, but memory space remains idle due to insufficient configurable processing units
Solution Approach 1:
The patent enables dynamic matching between processing units and memory resources across host boundaries. When processing units need to be allocated, the system dynamically identifies available memory spaces from any host and pairs them with available processing units, rather than being constrained by fixed host configurations. This dynamic approach ensures that memory resources are continuously utilized and can be quickly paired with processing units as capacity requirements change
4Ease of operation
If memory resources are strictly allocated to individual hosts, then host memory management is simplified, but hardware resource allocation flexibility deteriorates
Solution Approach 1:
The patent introduces a memory management intermediary layer that abstracts the complexity of cross-host memory allocation. This intermediary manages the mapping between virtual machine memory requests and physical host memory spaces, handling the complexity of resource tracking, allocation, and deallocation. By centralizing this management function, the system maintains operational simplicity for individual hosts while achieving high flexibility in overall resource allocation across the entire system
Data Source
AI summary
The present application discloses a computing system and a memory sharing method for a computing system. The computing system includes a first host, a second host, a first memory extension device and a second memory extension device. The first memory extension device is coupled to the first host. The second memory extension device is coupled to the second host and the first memory extension device. The first host accesses a memory space of a memory of the second host through the first memory extension device and the second memory extension device according to a first physical address of the first host.