Virtualization-Aware Overload Protection Engine for Network Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtual network functions (VNFs) in data centers often experience overload due to high-speed data plane and signaling processing, leading to congestion and unresponsiveness, which can cause bottlenecks in the data center infrastructure.
Innovation Solution
A virtualization-aware overload protection engine (OPE) is implemented in hardware, such as a network interface or fabric interface, to detect and mitigate overload conditions by prioritizing high-priority packets and discarding low-priority ones, preventing CPU overload and ensuring critical packets are forwarded while avoiding wholesale shutdown of queues and ports.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If VNF performs high-speed data plane and signaling processing, then processing capacity is improved, but CPU becomes overloaded and unresponsive
Solution Approach 1:
The overload protection engine performs preliminary actions by detecting overload conditions and prioritizing packets before they reach the CPU. It proactively identifies high-priority packets and prepares them for processing, preventing CPU overload before it occurs rather than reacting after the system becomes unresponsive.
Solution Approach 2:
The overload protection engine acts as an intermediary between the network interface and the CPU. It sits in the data path, inspecting packets, determining their priority, and managing the flow to the CPU, thereby protecting the CPU from being overwhelmed while maintaining high-speed processing capability.
2Measurement precision
If all incoming packets are forwarded to CPU for inspection, then packet handling accuracy is improved, but CPU resources are depleted
Solution Approach 1:
The overload protection engine applies local quality by treating different packets differently based on their priority. High-priority packets receive full inspection and forwarding to the CPU, while low-priority packets are handled differently (dropped or rate-limited), thereby optimizing CPU resource usage while maintaining accurate handling of critical packets.
Solution Approach 2:
Instead of applying full inspection to all packets (excessive action), the system applies partial action by selectively forwarding only high-priority packets to the CPU for detailed inspection. This partial approach maintains sufficient packet handling accuracy for critical traffic while significantly reducing CPU resource consumption.
3Reliability
If CPU discards packets or disables ports to prevent overload, then CPU protection is improved, but critical packets may be lost
Solution Approach 1:
The system performs preliminary action by identifying and prioritizing critical packets before the overload condition becomes severe. The overload protection engine proactively manages packet flow, ensuring critical packets are forwarded to the CPU even during overload conditions, thereby protecting the CPU while preventing loss of essential information.
Solution Approach 2:
The overload protection engine uses feedback mechanisms to monitor CPU load conditions and dynamically adjust packet handling. When overload is detected, it selectively forwards high-priority packets based on real-time conditions, providing feedback-driven protection that prevents CPU overload while minimizing critical packet loss through adaptive decision-making.
Data Source
AI summary
A fabric interface, including: an ingress port to receive incoming network traffic; a host interface to forward the incoming network traffic to a host; and a virtualization-aware overload protection engine including: an overload detector to detect an overload condition on the incoming network traffic; a packet inspector to inspect packets of the incoming network traffic; and a prioritizer to identify low priority packets to be dropped, and high priority packets to be forwarded to the host.