Transcendent Memory API for Distributed RAM Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed Shared Memory (DSM) systems face challenges such as slow data transfer between computer systems, granular remote access, and unpredictable performance due to varying types of RAM access, leading to slower performance compared to local hard disk access.
Innovation Solution
Implementing transcendent memory, a dynamic portion of RAM that allows multiple computer systems to efficiently share memory resources via a transcendent memory API, enabling data transfer and storage across systems for faster access times, using a network connection to store data in another computer's RAM instead of slower local storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data is transferred between computer systems using DSM, then storage capacity is improved, but access speed deteriorates
Solution Approach 1:
The patent segments memory access into two distinct paths: local memory access for high-speed operations and remote memory access via network for capacity expansion. This segmentation allows the system to maintain fast access speeds for frequently used data while utilizing remote storage capacity when needed, resolving the contradiction between storage capacity and access speed.
Solution Approach 2:
The patent introduces a memory management component that acts as an intermediary between applications and both local and remote memory systems. This intermediary automatically manages data placement and retrieval, transparently handling the complexity of distributed memory access while maintaining the illusion of unified memory to applications.
2Adaptability or versatility
If remote RAM is accessed at byte, word, or cache line level, then access flexibility is improved, but performance deteriorates
Solution Approach 1:
The patent implements page-level granularity for remote memory access rather than fine-grained byte or word-level access. This partial action approach provides sufficient flexibility for most applications while avoiding the performance penalty of excessive fine-grained access, achieving a practical balance between adaptability and productivity.
3Adaptability or versatility
If different types of RAM are used to access remote RAM, then memory compatibility is improved, but performance predictability deteriorates
Solution Approach 1:
The patent standardizes remote memory access operations to treat all remote memory uniformly, regardless of the physical memory type at the remote system. This homogenization approach ensures predictable performance by applying consistent access protocols and management strategies, eliminating the variability introduced by heterogeneous memory types.
4Quantity of substance
If data is stored on local hard disk, then storage capacity is improved, but access speed deteriorates
Solution Approach 1:
The patent implements a copying strategy where data is replicated between local and remote memory systems. Frequently accessed data is kept in local memory for fast access, while copies are maintained in remote memory for capacity expansion. This copying approach allows the system to achieve both fast access speeds and expanded storage capacity simultaneously.
Data Source
AI summary
Various arrangements for utilizing memory of a remote computer system are presented. Two computer systems may allocate a portion of RAM accessible to a memory-access API. A first set of data from the first portion of the first memory of a first computer system may be determined to be moved to memory of another computer system. The first set of data from the first portion of the first memory may be transmitted for storage in the second portion of the second memory of a second computer system. Using the second memory-access API, the set of data may be stored in the second portion of the second memory. Using the first memory-access API, the set of data from the first portion of the first memory may be deleted.