Network Egress Link Buffer Management via Packet Marking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network systems face challenges in efficiently managing packet flow across interconnects, leading to buffer overflow and redundant traffic, which reduces bandwidth utilization and increases costs due to the need for larger buffer memory and power consumption.
Innovation Solution
The method involves a system where a source network transmits packets with preferential treatment indicators, allowing edge devices to determine packet loss and adjust flow rates without modifying the overall flow rate, thereby scaling back problematic flows and reducing redundant traffic on interconnects, allowing smaller buffer sizes while maintaining bandwidth utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network devices use larger buffer memory to prevent packet loss, then packet transmission reliability is improved, but device cost and power consumption increase
Solution Approach 1:
The patent segments packet flows by marking certain packets with preferential treatment indicators while leaving others unmarked. This segmentation allows the system to differentiate between packets that require reliable delivery and those that can tolerate loss, enabling selective buffering strategies that reduce overall buffer requirements while maintaining reliability for critical packets.
Solution Approach 2:
The patent applies local quality by providing different treatment to different packets within the same flow based on their marking status. Marked packets receive preferential treatment including priority buffering and transmission, while unmarked packets receive standard treatment. This localized differentiation optimizes buffer usage by concentrating resources on packets where reliability is most critical.
2Reliability
If network devices use larger buffer memory to handle packet flow, then packet loss is reduced, but power consumption increases
Solution Approach 1:
The patent segments the packet flow into marked and unmarked packets, applying different buffering and transmission strategies to each segment. This segmentation enables the system to maintain reliable delivery for marked packets using targeted buffering, while reducing or eliminating buffering for unmarked packets, thereby reducing overall power consumption associated with buffer management.
Solution Approach 2:
The patent changes the buffering parameter dynamically based on packet marking status. Instead of using a fixed large buffer for all packets, the system adjusts buffer allocation and retention time based on whether a packet is marked for preferential treatment. This parameter change optimizes the balance between reliability and power consumption.
3Productivity
If network devices increase buffer size to maintain high utilization, then bandwidth utilization is maintained, but device complexity increases
Solution Approach 1:
The patent simplifies buffer management by segmenting packets into marked and unmarked categories with distinct handling rules. Marked packets are buffered and prioritized, while unmarked packets are transmitted immediately without buffering. This segmentation reduces the complexity of buffer management compared to traditional approaches that require complex algorithms to manage large buffers for all packets.
Solution Approach 2:
The patent applies local quality by implementing different buffer management policies for marked versus unmarked packets. This localized approach simplifies overall system complexity by replacing complex global buffer management with simple, rule-based local decisions at each network device.
4Quantity of substance
If network devices use small buffers to reduce cost, then device cost is reduced, but packet loss increases leading to redundant traffic
Solution Approach 1:
The patent segments packet flows to identify marked packets that require reliable delivery. By segmenting the flow, the system can use small buffers selectively for marked packets while transmitting unmarked packets immediately, thereby reducing overall buffer requirements while minimizing redundant traffic through targeted retransmission only when necessary.
Solution Approach 2:
The patent implements feedback mechanisms where receiving devices send acknowledgments for marked packets. This feedback enables the sending device to detect packet loss and trigger selective retransmission, reducing redundant traffic by retransmitting only lost marked packets rather than entire flows, thereby optimizing the trade-off between small buffer size and traffic efficiency.
Data Source
AI summary
Methods and systems for managing packet flow in a local network in a manner that allows network devices at the edge of the network to use minimal buffer memory while maximizing use of bandwidth allocated on an interconnect between the local network and an external network. Packet flows facing external network problems are scaled back, reducing redundant traffic on the interconnect and allowing for use of small buffers in edge devices. A flow source marks a subset of packets within a flow for preferential treatment within the local network. The flow source then adjusts the flow rate only responsive to failures in transmission of the marked packets. In some implementations, an edge device removes the markings prior to packet egress. The local network honors the preferential treatment markings such that a loss of a marked packet is more likely to occur on an external network than on the local network.