Smart NIC Workload Placement for 5G Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Software-based network functions in telecommunications, particularly in 5G deployments, face bottlenecks in throughput and latency due to increasing virtualized network functions, which are critical for applications like remote surgery and autonomous vehicles.
Innovation Solution
A network interface controller-based orchestration platform that offloads latency-sensitive virtualized workloads to smart-NICs, leveraging usage metrics to optimize workload placement across nodes in a cloud infrastructure, thereby reducing latency and maximizing throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If more virtualized network functions are deployed on cloud architectures, then service functionality and adaptability are improved, but throughput and latency performance deteriorate due to processor bottlenecks
Solution Approach 1:
The patent segments the network function processing by separating control plane functions (remaining on CPU) from data plane functions (offloaded to NIC). This segmentation allows the system to maintain service adaptability while improving throughput by parallelizing processing across different hardware components with specialized functions.
Solution Approach 2:
The patent introduces a network interface controller with integrated service gateway function as an intermediary between the cloud infrastructure and network traffic. This intermediary handles data plane processing, relieving the CPU bottleneck and improving overall system throughput while maintaining service functionality.
2Ease of operation
If virtualized network functions are executed on commodity processors, then deployment ease and adaptability are improved, but processing speed and latency are worsened due to end of Moore's law
Solution Approach 1:
The patent replaces the general-purpose mechanical processing system (CPU) with a specialized electronic processing system (NIC with integrated service gateway). This substitution leverages the NIC's hardware-accelerated packet processing capabilities to achieve higher speeds while maintaining deployment simplicity through virtualization.
Solution Approach 2:
The patent changes the processing parameter from software-based CPU execution to hardware-based NIC execution. This parameter change exploits the NIC's dedicated network processing architecture, providing faster processing speeds for network-intensive operations while preserving the flexibility of virtualized deployment.
3Reliability
If network functions are hardware-based proprietary devices, then processing performance and reliability are improved, but deployment complexity and adaptability are worsened
Solution Approach 1:
The patent makes the NIC universal by integrating multiple functions (network interface, service gateway, packet processing) into a single hardware component. This multi-functionality allows the system to achieve hardware-level reliability and performance while simplifying deployment through a standardized, multi-purpose platform that can handle various network functions.
Solution Approach 2:
The patent merges previously separate hardware functions (network interface card and service gateway) into a single integrated NIC architecture. This consolidation maintains the reliability of hardware-based processing while reducing deployment complexity by eliminating the need for separate proprietary hardware devices.
Data Source
AI summary
A computer program product comprises a computer readable storage medium having program instructions embodied therewith. The program instructions are executable by one or more processors to cause the one or more processors to collect, at a network interface controller associated with a first node, a set of usage metrics from one or more other network interface controllers respectively associated with one or more other nodes. Further, the program instructions are executable by one or more processors to cause the one or more processors to select, at the network interface controller associated with the first node, one of the one or more other nodes to allocate a virtualized workload for execution, wherein the selection utilizes at least a portion of the set of usage metrics collected from each of the one or more other network interface controllers.


