Disaggregation Controller Memory Access Delay Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face challenges in efficiently managing and accessing memory resources across heterogeneous computing devices connected via networks, particularly due to increased data-intensive workloads and physical positional limitations, leading to performance degradation and response delays.
Innovation Solution
A disaggregation computing system is implemented, comprising local and remote computing devices with controllers and a network that checks and controls memory access delays, enabling adaptive management of memory access and addressing to expand processing capacity and secure performance by directly accessing remote disaggregated memory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If memory is disaggregated across remote computing devices connected via network, then memory capacity and scalability are improved, but access response delay and network latency increase
Solution Approach 1:
The system performs preliminary actions by predicting future memory access patterns and pre-fetching data from remote disaggregated memory before actual access is needed. The prediction unit analyzes access patterns and pre-loads data into local memory or cache, thereby reducing the effective access delay when data is actually needed, while still maintaining the benefit of remote memory capacity.
Solution Approach 2:
The patent introduces intermediate memory structures (local memory, cache memory) between the processor and remote disaggregated memory. These intermediaries buffer and manage data transfer, reducing the impact of network latency by keeping frequently accessed data locally while still utilizing remote memory capacity for large-scale storage.
2Adaptability or versatility
If heterogeneous computing devices share remote memory via network, then resource utilization and adaptability are improved, but system complexity and management difficulty increase
Solution Approach 1:
The patent implements a universal memory management architecture where the disaggregation controller and prediction unit handle multiple types of computing devices (CPU, GPU, accelerator) through a unified interface and protocol. This multi-functional approach allows heterogeneous devices to access remote memory seamlessly without requiring device-specific management logic, thereby reducing system complexity while maintaining high resource utilization.
Solution Approach 2:
The system employs feedback mechanisms where the prediction unit continuously monitors memory access patterns from heterogeneous devices and adjusts pre-fetching strategies accordingly. This closed-loop control optimizes resource allocation dynamically, allowing the system to adapt to different workload types and device characteristics without manual configuration, thus reducing management complexity.
3Productivity
If remote memory access is enabled to overcome physical limitations, then processing capacity is improved, but network delay and performance degradation occur
Solution Approach 1:
The prediction unit performs preliminary analysis of memory access patterns and pre-fetches data before actual processing needs arise. This allows the system to maintain high processing capacity by ensuring data is available locally when needed, while still utilizing remote memory for expanded capacity, thereby minimizing the impact of network delay on overall productivity.
Data Source
AI summary
Disclosed herein a disaggregation computing system. The disaggregation computing system comprising: a local computing device that comprises a local processor, a local memory bus, a local memory and a local disaggregation controller; a remote computing device that comprises a remote processor, a remote memory bus, a remote memory and a remote disaggregation controller; and a disaggregation network that connects the local computing device and the remote computing device, wherein the local disaggregation controller and the remote disaggregation controller are configured to: check a response delay for access of the remote memory, and control the access of the remote memory based on the response delay.