Persistent Memory Storage Nodes Without CPUs

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current data storage systems in datacenters using persistent memory (PM) are costly due to the need for storage nodes with CPUs, and there is a lack of approaches to build PM-based systems that avoid these high costs while maintaining performance.

Innovation Solution

A method for building a PM-based data storage system without processors at storage nodes, where data is stored in PM and accessed directly by remote compute nodes through a network, with metadata managed by global controllers, and data is stored in link lists, allowing compute nodes to request and write data directly to PM using RDMA operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If storage nodes are equipped with CPUs to manage and process data storage operations, then data processing capability and system control are improved, but hardware cost and energy consumption increase significantly

Engineering Contradiction:

Improvedata processing capabilityVSAvoidhardware cost

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The patent extracts the CPU component from storage nodes, creating 'dumb' storage nodes that contain only persistent memory without processing units. This eliminates the need for expensive CPU hardware at storage nodes while maintaining data storage functionality. The processing capabilities are extracted and consolidated at compute nodes instead.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces an intermediary mechanism where compute nodes act as mediators between applications and storage nodes. Compute nodes handle all processing and control operations, communicating with simplified storage nodes through standardized interfaces. This intermediary approach allows complex operations to be performed without requiring CPUs at storage nodes.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If storage nodes are equipped with CPUs to manage data operations, then system control and data management capability are improved, but energy consumption increases due to running CPUs

Engineering Contradiction:

Improvesystem control capabilityVSAvoidenergy cost of running CPUs

Core Design Contradiction:

ReliabilityVSUse of energy by stationary object

Solution Approach 1:

The patent removes CPUs from storage nodes, eliminating the energy consumption associated with running processor units at storage locations. Only essential memory management functions remain, which consume minimal energy compared to full CPU operations.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent implements self-service mechanisms where storage nodes autonomously manage their own memory operations without requiring external CPU control. Data is organized in link lists that can be traversed and managed through simple pointer operations, enabling storage nodes to serve themselves without active CPU intervention.

Inventive Principle:

Principle #25Self-service

3Quantity of substance

If data is stored in traditional hard disks or SSDs, then storage capacity is achieved, but data access speed is slow

Engineering Contradiction:

Improvestorage capacityVSAvoiddata access speed

Core Design Contradiction:

Quantity of substanceVSSpeed

Solution Approach 1:

The patent changes the physical parameter of storage media from traditional hard disks or SSDs to persistent memory. This parameter change provides both large storage capacity and fast access speeds, as persistent memory offers nanosecond-to-microsecond access times compared to the millisecond access times of traditional storage devices.

Inventive Principle:

Principle #35Parameter changes

4Productivity

If compute nodes access data through storage node CPUs, then data processing is enabled, but network traffic and processing overhead increase

Engineering Contradiction:

Improvedata processing throughputVSAvoidnetwork communication overhead

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent extracts the CPU layer from storage nodes, eliminating the intermediate processing step that caused network overhead. Compute nodes can directly access persistent memory at storage nodes without routing through CPU layers, reducing network communication steps and processing overhead.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS11611617B2Distributed data store with persistent memory

Publication Date: 2023.03.21 PURDUE RES FOUND

AI summary

A method to build a persistent memory (PM)-based data storage system without involving a processor (CPU) at storage nodes is disclosed which includes storing data in one or more storage nodes that only include PM and no CPUs, with data stored in PM in form of link lists, accessing data stored in the one or more storage nodes' PM directly by remote compute nodes through a network, maintaining metadata associated with the data by one or more global controllers (metadata servers), upon request by a user to read or write data, the compute nodes contacting the one or more metadata servers to obtain location of data of interest in form of pointers (shortcuts), and the compute nodes sending network requests directly to the one or more storage nodes' PM to locate latest version of data by tracing the link list from the associated shortcut to corresponding tails.