Cluster Memory Integrated Management for VM Capacity Expansion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory expansion technologies face limitations in capacity expansion due to DRAM limitations, leading to performance fluctuations and high costs, as well as slower access speeds of non-volatile memory, which affect application performance in cluster systems.
Innovation Solution
A method and apparatus for memory integrated management in a cluster system that allocates high-performance DRAM and high-integration memory across multiple physical nodes to maximize throughput by profiling memory access patterns and distributing memory resources efficiently, ensuring optimal performance and capacity utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If memory expansion technology is used to increase total memory capacity, then the memory capacity available to virtual machines is improved, but application performance fluctuates greatly and access speed decreases
Solution Approach 1:
The patent segments memory resources by type (local DRAM vs. remote memory) and allocates them through separate distributors. The local DRAM distributor manages high-speed memory within each physical node, while the remote memory distributor manages expanded memory across the cluster, allowing each type to be optimized independently for its intended use case
Solution Approach 2:
The patent introduces memory profilers as intermediary components that analyze memory access patterns and provide feedback to the memory distributors. These profilers act as mediators between applications and the expanded memory system, predicting performance changes and enabling the distributors to make intelligent allocation decisions that maintain application performance while expanding capacity
2Quantity of substance
If DRAM from other computers is connected for memory expansion, then memory capacity is increased, but the cost of memory expansion becomes expensive
Solution Approach 1:
The patent creates a universal memory pool that aggregates DRAM resources from multiple physical nodes in the cluster. This pooled memory can be dynamically allocated to any virtual machine on any node, allowing existing memory resources to serve multiple purposes and reducing the need for additional dedicated memory purchases
Solution Approach 2:
The patent implements a memory recycling mechanism where unused or underutilized memory resources from one physical node can be allocated to virtual machines on other nodes. The memory pool distributor continuously monitors and redistributes memory resources, ensuring that idle memory capacity is recovered and reused rather than remaining wasted
3Quantity of substance
If non-volatile memory or high-performance SSD is used for memory expansion, then memory capacity is increased, but access speed becomes tens of times slower than DRAM
Solution Approach 1:
The patent applies local quality by ensuring that each physical node retains its local DRAM with high access speed for time-critical operations, while using remote memory (non-volatile memory or SSD from other nodes) only for capacity expansion. The system recognizes that different memory locations have different quality characteristics and allocates workloads accordingly based on access speed requirements
4Quantity of substance
If memory expansion is implemented, then total memory capacity is increased, but the rate of application access to expanded memory causes performance degradation
Solution Approach 1:
The patent implements feedback through memory profilers that continuously monitor memory access patterns and predict performance changes. This feedback loop allows the memory distributors to adjust allocations in real-time, preventing performance degradation by identifying and addressing access pattern issues before they significantly impact application reliability
Data Source
AI summary
An apparatus for memory integrated management in a cluster system including a plurality of physical nodes connected to each other by a network determines one of the plurality of physical nodes as a node to place a new virtual machine, allocates the first type of memory allocated to the one physical node to the new virtual machine as much as the memory capacity required by the new virtual machine, and distributes the second type of memory to a plurality of virtual machines running on the plurality of physical nodes by integrating and managing the second type of memory allocated to each of the plurality of physical nodes. In this case, the access speed of the second type of memory is faster than that of the first type of memory.