DPU Offload for 5G Edge Compute and Network Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data center edge sites face challenges in efficiently utilizing computing and networking resources due to physical space and power limitations, with high costs associated with server and switch infrastructure, and the need to balance cloud network management with user applications.
Innovation Solution
Disaggregate and offload cloud network management functions to hardware-based network devices like DPUs, allowing these functions to be performed without software-based processing on virtual machines, thereby optimizing CPU capacity and reducing server requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If cloud network management functions are performed on general-purpose servers using virtual machines, then service flexibility and scalability are improved, but resource consumption (CPU cycles, memory, power) and cost increase
Solution Approach 1:
The patent segments cloud network management functions from general-purpose server workloads by introducing specialized network devices (DPUs, sNICs) that handle networking tasks independently. This segmentation allows servers to focus on compute-intensive user applications while network management is offloaded to dedicated hardware, reducing resource consumption without sacrificing service flexibility.
Solution Approach 2:
The patent introduces network devices with intelligent agents as intermediaries between servers and the network infrastructure. These agents handle network management functions (traffic steering, load balancing, security policies) without requiring server CPU involvement, thus reducing resource consumption while maintaining adaptable network control.
2Reliability
If more servers are deployed at edge sites to handle both user applications and network management, then service capacity and availability are improved, but physical space requirements and infrastructure cost increase
Solution Approach 1:
The patent extracts network management functions from general-purpose servers and places them on specialized network devices at the edge site. This extraction allows the same physical infrastructure to support both user applications and network management without requiring additional servers, thereby maintaining service availability while reducing physical space requirements.
Solution Approach 2:
The patent makes network devices multi-functional by enabling them to handle both traditional networking tasks and cloud network management functions (traffic steering, load balancing, security). This universality allows a single device to perform multiple roles that previously required separate servers, reducing the physical footprint while maintaining service capacity and availability.
3Ease of operation
If CPU capacity is allocated to virtual machines for network management tasks, then network control and orchestration are improved, but capacity available for user applications decreases
Solution Approach 1:
The patent introduces intelligent agents on network devices as intermediaries that handle network control and orchestration tasks. These agents communicate with the cloud management plane and enforce policies locally without consuming server CPU capacity, thus maintaining ease of network operation while preserving full CPU capacity for user applications.
Solution Approach 2:
The patent replaces software-based network management processing on general-purpose CPUs with hardware-accelerated processing on specialized network devices. This substitution moves network control functions from the software layer (consuming CPU cycles) to the hardware layer, maintaining network control capabilities while freeing up CPU capacity for high-performance application processing.
Data Source
AI summary
A digital processing unit (DPU) is configured to disaggregate computing service provider functions of a cloud service provider from hosts of an edge computing network. The hosts are implemented on servers hosting a plurality of virtual machines or containers. The edge computing network comprises computing and storage devices configured to extend computing resources of the cloud service provider to remote users of the cloud service provider at a location remote from the cloud service provider. The DPU executes a computing service provider function that is disaggregated from processing cores of the server.


