Edge NIC Virtual Platform Latency Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Edge computing solutions face inefficiencies due to physical distance between data computation devices and data generating devices, leading to increased latency, and inflexibility in responding to varying workload demands, particularly in network-bound workloads and multi-tenant scenarios.
Innovation Solution
The implementation of smart Network Interface Cards (NICs) with logic to divide into virtual partitions, providing programmable compute capabilities closer to requesting devices, enabling intelligent resource allocation and management, and allowing for zero-touch provisioning and secure authentication without relying on external servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If data computation devices are physically separated from data generating devices, then network infrastructure can be simplified, but latency increases
Solution Approach 1:
The patent segments the network infrastructure into distributed Edge computing nodes that process data locally near data generating devices, rather than using a centralized cloud architecture. This segmentation allows computation to occur at multiple distributed points, reducing the physical distance data must travel and thereby reducing latency while maintaining infrastructure simplicity.
Solution Approach 2:
The patent introduces a spatial dimension to computing by distributing Edge nodes across multiple geographic locations closer to data sources. Instead of a single centralized cloud datacenter, the system creates a multi-dimensional network of computation points that can simultaneously process data from different sources, reducing transmission distance and latency.
2Loss of time
If Edge computing infrastructure is deployed closer to endpoint devices, then latency is reduced, but system flexibility and adaptability decrease
Solution Approach 1:
The patent implements dynamic resource allocation and workload management in Edge computing nodes, allowing the system to adapt to varying workload demands in real-time. The Edge infrastructure can dynamically provision computing resources, adjust service configurations, and respond to changing conditions, thereby maintaining flexibility and adaptability despite being distributed closer to endpoints.
Solution Approach 2:
The Edge computing nodes are designed with multi-functional capabilities to handle diverse workload types including network function virtualization, content delivery, data processing, and service orchestration. This universality allows a single Edge infrastructure to adapt to multiple different workloads and service requirements, maintaining versatility across various application scenarios.
3Device complexity
If traditional cloud infrastructure is used, then resource management is centralized and simple, but scalability and responsiveness to varying workloads are limited
Solution Approach 1:
The patent segments the centralized cloud resource management into distributed Edge computing nodes that can independently manage local resources while coordinating with the broader network. This segmentation enables parallel resource management across multiple nodes, significantly improving scalability and the ability to respond to varying workload demands without overwhelming a single centralized management system.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods, apparatus, systems, and articles of manufacture are disclosed to facilitate network interface device based edge computing. A disclosed example includes threshold analyzer circuitry to identify a resource inactivity threshold for first a resource associated with the network interface device, the resource inactivity threshold indicative of a duration of time that the first resource is not utilized; resource allocator circuitry to determine that the resource meets the resource inactivity threshold; and virtual platform manager circuitry to generate a virtual platform associated with the network interface device based on the determination, the generated virtual platform to reduce latency in execution of a function.