Datacenter Congestion Control via RTT Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing congestion management techniques in datacenter networks fail to distinguish between fabric and engine congestion, leading to inadequate traffic management, particularly in cloud computing environments where network devices handle multiple virtualized or containerized environments.
Innovation Solution
The system employs a method to differentiate between fabric and engine congestion by measuring and managing the congestion window size based on round-trip time (RTT) components, specifically calculating engine RTT and fabric RTT to adjust congestion window sizes accordingly, allowing for targeted congestion management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional congestion management techniques are used, then network devices can handle basic traffic flow, but they fail to distinguish between fabric congestion and engine congestion, leading to inadequate traffic management
Solution Approach 1:
The patent segments congestion identification into two distinct components: fabric RTT measurement for network path congestion and engine RTT measurement for processing congestion. This segmentation allows the system to precisely identify the source of congestion by measuring round-trip times separately for fabric transmissions and engine processing, thereby resolving the technical contradiction between measurement precision and device complexity.
2Reliability
If congestion window size is reduced to manage congestion, then packet loss is controlled, but network throughput decreases
Solution Approach 1:
The patent implements dynamic congestion window adjustment based on the type of congestion detected. When fabric congestion is detected through RTT measurement, the congestion window is adjusted differently than when engine congestion is detected. This dynamic adjustment strategy allows the system to maintain packet loss control while maximizing network throughput by adapting the congestion window size to the specific congestion condition.
3Productivity
If congestion window size is increased to improve throughput, then network productivity increases, but congestion control becomes inadequate
Solution Approach 1:
The patent employs feedback mechanisms where RTT measurements continuously monitor network conditions and engine processing status. Based on this feedback, the system dynamically adjusts the congestion window size to maintain optimal throughput while preventing congestion. The feedback loop ensures that when congestion is detected, the congestion window is adjusted appropriately, thereby maintaining both high productivity and reliable congestion control.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
Systems and methods for controlling congestion of a data network are provided. An engine roundtrip time (RTT) and a fabric RTT for a network flow are determined. An engine-based congestion window size for the flow is determined based on the engine RTT and a target engine RTT. A fabric-based congestion window size for the flow is determined based on the fabric RTT and a target fabric RTT. The smaller of the engine-based congestion window size and the fabric-based window size is selected for use in transmitting a future packet associated with the flow. The target engine RTT is determined based in part on the current congestion window used to transmit packets for the flow and/or the target fabric RTT is determined based on a number of hops packets associated with the flow traverse from a source to a destination associated with the flow.