Rack Switch Packet Offload for Server CPU Workload Relief
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing burden on server processors due to high bit-rate network traffic and complex packet processing activities in data centers, particularly for east-west traffic, leads to inefficient utilization of CPU resources, which could be used for more revenue-generating tasks.
Innovation Solution
A switch with infrastructure offload capabilities, including CPUs and accelerators, performs packet processing and network termination, freeing server CPUs for other tasks, and manages memory transactions and virtualized execution environments, reducing the need for server-based processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If server processors handle packet processing activities, then network traffic can be processed, but server CPU resources are consumed that could be used for revenue-generating tasks
Solution Approach 1:
The patent extracts packet processing functions from server processors and relocates them to network switches. The switch is enhanced with processing capabilities including parsers, de-encapsulation engines, and decryption modules that previously resided on servers. This extraction frees server CPU resources for revenue-generating workloads while maintaining network processing capabilities.
Solution Approach 2:
The network switch acts as an intermediary between network traffic and server processors. Instead of servers directly handling packet processing, the switch intermediates by performing initial packet processing, filtering, and forwarding only relevant traffic to servers. This reduces the burden on server CPUs while maintaining system functionality.
2Reliability
If servers perform network processing, then packets can be processed, but the burden on servers increases with increased packet volumes and processing complexity
Solution Approach 1:
The patent segments network processing functions into distinct modules distributed across switches and servers. The switch handles packet parsing, de-encapsulation, decryption, and filtering functions, while servers focus on application-level processing. This segmentation reduces server complexity and distributes the processing burden appropriately.
Solution Approach 2:
The patent replaces traditional server-based packet processing mechanics with specialized network switch hardware. The switch incorporates dedicated processing engines and accelerators that are optimized for network packet processing, substituting the general-purpose server processing mechanism with a specialized system better suited for high-volume packet handling.
3Adaptability or versatility
If CPUs are used for packet processing, then network traffic handling is enabled, but fewer CPU resources remain for other billable services
Solution Approach 1:
The patent extracts network processing responsibilities from server CPUs and assigns them to network switches. This extraction enables server CPUs to be reallocated to billable services such as cloud computing workloads, virtualization services, and application hosting, thereby increasing service portfolio flexibility and productivity.
Data Source
AI summary
Examples described herein relate to a switch device for a rack of two or more physical servers, wherein the switch device is coupled to the two or more physical servers and the switch device performs packet protocol processing termination for received packets and provides payload data from the received packets without a received packet header to a destination buffer of a destination physical server in the rack. In some examples, the switch device comprises at least one central processing unit, the at least one central processing unit is to execute packet processing operations on the received packets. In some examples, a physical server executes at least one virtualized execution environments (VEE) and the at least one central processing unit executes a VEE for packet processing of packets with data to be accessed by the physical server that executes the VEE.