Network Data Flow Management via Protocol-Specific Packet Dropping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network congestion management techniques, such as tail drop and random early detect (RED), often penalize short-lived small data flows and fail to efficiently handle newer protocols like UDP, leading to poor network performance and increased processor and memory utilization.
Innovation Solution
A system and method that selectively apply packet treatment labels to data packets based on their protocols, dropping packets from congested queues according to pre-defined criteria, including TCP window size and counter values, to prioritize data flows and manage congestion effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If tail drop is used to relieve network congestion, then network congestion is reduced, but network performance for the majority of users deteriorates
Solution Approach 1:
The patent applies different drop criteria to different data flows based on their protocol type. TCP flows use window-size-based criteria while UDP flows use different criteria, creating localized quality differences in packet treatment that optimize overall network performance while managing congestion
2Quantity of substance
If random early detect (RED) is used to drop packets before hard limit, then network congestion is managed, but processor and memory utilization increases
Solution Approach 1:
The patent pre-calculates and stores drop criteria for different protocol types before congestion occurs. During congestion, the router simply applies the pre-determined criteria rather than performing complex real-time calculations, reducing processor and memory utilization while effectively managing congestion
3Quantity of substance
If RED drops packets based on statistical methods, then network congestion is managed, but short-lived small data flows are penalized
Solution Approach 1:
The patent changes the dropping parameter from random statistical selection to protocol-specific criteria based on TCP window size for TCP flows and different criteria for UDP flows. This parameter change ensures that packet drops are more fair and less likely to penalize short-lived small data flows while still managing congestion effectively
Data Source
AI summary
The present disclosure is directed to a system and method to manage data flow in a network. The method includes receiving a plurality of data flows, wherein each of the plurality of data flows includes a plurality of data packets. The method also includes determining a protocol associated with each of the plurality of data flows. The method also includes dropping at least one data packet from at least one data flow associated with a first protocol, wherein the at least one dropped data packet meets a first criterion associated with the first protocol. The method also includes dropping at least one additional data packet from a data flow associated with a second protocol, wherein the at least one additional data packet meets a second criterion associated with the second protocol.


