Multi-level iSCSI QoS for DCB Network Traffic Differentiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Data Center Bridging (DCB) networks, existing technologies fail to differentiate between high and low priority iSCSI storage traffic effectively, leading to competition for network resources and potential degradation of response times for high priority traffic during congestion, resulting in starvation and high latencies.
Innovation Solution
Implementing a method to sub-classify iSCSI traffic based on internal classifiers, directing packets into distinct channels with different priority levels, bandwidth allocations, and lossless or lossy configurations using the DCB protocol, allowing for dynamic QoS management within the same traffic type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If iSCSI traffic is prioritized as a single traffic class with uniform QoS properties, then implementation simplicity is maintained, but high priority storage traffic cannot be differentiated from low priority traffic, leading to resource competition and latency degradation
Solution Approach 1:
The patent segments iSCSI traffic into multiple priority classes (e.g., priority 4 for high priority storage traffic, priority 5 for low priority replication traffic) within the DCB framework. This segmentation allows differentiated QoS treatment for different types of storage traffic while maintaining compatibility with existing DCB infrastructure, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The patent applies local quality by assigning different QoS properties (bandwidth allocation, losslessness, priority levels) to different iSCSI traffic classes based on their specific requirements. High priority storage traffic receives lossless channels with guaranteed bandwidth, while low priority replication traffic can use lossy channels, optimizing resources for each traffic type's local needs.
2Reliability
If all iSCSI traffic is routed through lossless channels with flow control, then data integrity is ensured, but network congestion increases and latency increases for non-critical traffic
Solution Approach 1:
The patent applies local quality by matching channel characteristics to traffic requirements: critical storage traffic is routed through lossless channels with flow control to ensure data integrity, while non-critical replication traffic is routed through lossy channels that allow congestion and frame discards, thereby reducing latency for acceptable data loss scenarios.
Solution Approach 2:
The patent changes the QoS parameters (losslessness, bandwidth allocation, priority level) based on traffic type. By dynamically adjusting these parameters for different iSCSI traffic classes, the system achieves reliable transmission for critical data while minimizing latency for non-critical data, resolving the contradiction between reliability and time loss.
3Productivity
If bandwidth is allocated uniformly across all iSCSI traffic, then resource management is simplified, but high priority traffic experiences starvation during network congestion
Solution Approach 1:
The patent segments bandwidth allocation by creating dedicated bandwidth pools for different iSCSI priority classes. High priority storage traffic is allocated guaranteed bandwidth in lossless channels, while low priority replication traffic receives remaining bandwidth in lossy channels. This segmentation ensures critical applications maintain productivity during congestion while managing bandwidth through structured DCB configurations.
4Ease of operation
If single priority channel is used for iSCSI traffic, then configuration and management is simplified, but HOL blocking occurs and response times degrade under congestion
Solution Approach 1:
The patent segments iSCSI traffic into multiple priority channels (priority 4 and priority 5) within the DCB framework, allowing high priority storage traffic and low priority replication traffic to be transmitted simultaneously without mutual blocking. This segmentation prevents HOL blocking by providing separate transmission paths, maintaining ease of operation through DCB's native multi-priority support while improving response times for critical traffic.
Data Source
AI summary
In an embodiment, a method can include setting metadata of a first data packet to have a first priority level based on a first internal classifier of the first data packet and setting metadata of a second data packet to have a second priority level based on a second internal classifier of the second data packet. The first data packet and second data packet can be of the same traffic type (e.g., iSCSI). The method can further include sub-classifying (e.g., the same traffic type), based on the first and second priority levels, to direct the first data packet into a first data channel and second data packet into a second data channel for the same traffic type.