Disaggregated Computing Memory Bandwidth Utilization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Cloud computing faces inflexibility in resource configuration and allocation, leading to poor resource utilization and high costs due to fixed hardware configurations and limitations in scaling or descaling demands, which restricts the ability to respond effectively to varying workload requirements.

Innovation Solution

A disaggregated computing system with dynamic resource allocation allows for flexible assembly of computing resources, where memory devices are assigned to processor devices, enabling the use of underutilized memory bandwidth for analytic functions, thereby optimizing memory bandwidth and improving resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If fixed hardware configurations are used in cloud computing, then resource allocation is simplified, but resource utilization efficiency deteriorates

Engineering Contradiction:

Improveresource allocation complexityVSAvoidresource utilization efficiency

Core Design Contradiction:

Device complexityVSProductivity

Solution Approach 1:

The system segments computing resources into separate pools: compute pools containing processor devices and memory pools containing memory devices. These segmented pools can be independently managed and dynamically allocated to different workloads, enabling flexible resource composition without fixed hardware configurations and thereby improving resource utilization efficiency.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system implements dynamic resource allocation where the configuration of computing resources can change over time based on workload demands. Memory bandwidth is dynamically allocated between primary compute tasks and analytic functions, and compute pools can be scaled up or down to match varying workload requirements, resolving the contradiction between allocation simplicity and utilization efficiency.

Inventive Principle:

Principle #15Dynamics

2Productivity

If memory bandwidth is fully allocated to primary compute tasks, then compute performance is maximized, but opportunistic analytic processing capability deteriorates

Engineering Contradiction:

Improvecompute performanceVSAvoidanalytic processing capability

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system enables continuous useful action by utilizing memory bandwidth that would otherwise be idle during primary compute tasks for additional analytic processing. This allows the system to continuously perform both primary compute functions and opportunistic analytic functions without sacrificing either, maximizing overall resource utility and maintaining adaptability for diverse workloads.

Inventive Principle:

Principle #20Continuity of useful action

Solution Approach 2:

The memory subsystem serves multiple functions simultaneously: it supports primary compute tasks while also enabling opportunistic analytic processing on resident data. This multi-functionality allows the same memory infrastructure to serve different purposes at different times, improving both compute performance and analytic processing capability without requiring separate dedicated resources.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Stability of the object's composition

If resources are statically allocated, then system stability is improved, but responsiveness to varying workload demands deteriorates

Engineering Contradiction:

Improvesystem stabilityVSAvoidworkload responsiveness

Core Design Contradiction:

Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The system transitions from static to dynamic resource allocation, allowing resource composition to adapt to varying workload demands while maintaining system stability through controlled management. The disaggregated architecture enables compute pools and memory pools to be dynamically scaled and reconfigured based on real-time workload conditions, resolving the contradiction between stability and responsiveness.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes key parameters of resource allocation dynamically: memory bandwidth allocation ratios, compute pool sizes, and memory pool capacities are adjusted based on workload characteristics. This parameter flexibility allows the system to maintain stability through controlled changes while responding effectively to varying workload demands, unlike rigid static allocation.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11474697B2Opportunistic data analytics using memory bandwidth in disaggregated computing systems

Publication Date: 2022.10.18 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Various embodiments for optimizing memory bandwidth in a disaggregated computing system, by a processor device, are provided. Respective memory devices are assigned to respective processor devices in the disaggregated computing system, the disaggregated computing system having at least a pool of the memory devices and a pool of the processor devices. An analytic function is performed on data resident in the pool of the memory devices using memory bandwidth not currently committed to a primary compute task.