Storage Network Compression Switching for Congestion Relief
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face network congestion due to explosive data growth and increasing expectations for availability, leading to higher latency and I/O failures, as multiple hosts access heavy data simultaneously, causing performance issues throughout the storage network.
Innovation Solution
A method that dynamically switches data access mode from normal to compressed mode in identified data links, using available CPU resources to increase networking device resources and alleviate congestion without affecting application data transmission, allowing for high endurance and robustness in storage systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is transmitted in normal (uncompressed) mode, then data transmission speed is maintained, but network congestion occurs and available resources are insufficient
Solution Approach 1:
The system dynamically switches between normal data access mode and compressed data access mode based on real-time network conditions. When network congestion is detected, the system transitions to compressed mode for affected data links, and switches back to normal mode when congestion is resolved, making the system adaptable to varying network loads
Solution Approach 2:
The system changes the data compression parameter from 0 (uncompressed) to 1 (compressed) based on network congestion conditions. This parameter change allows the same data link to operate in different compression states, effectively managing network resources during high-load periods while maintaining normal operation during low-load periods
2Reliability
If data is compressed, then available network resources increase, but additional processing overhead is required
Solution Approach 1:
The storage devices themselves perform the compression and decompression operations using their own available CPU resources. The networking device detects congestion and identifies which storage devices have sufficient processing capacity, then directs compressed data transmission through those devices, making them self-serve their own compression needs without requiring external compression hardware
Solution Approach 2:
The system applies compression selectively to specific data links based on local conditions at each storage device. Not all storage devices are compressed simultaneously - only those with sufficient available CPU resources are selected for compression, allowing the system to distribute the compression load across multiple devices with varying processing capacities
3Reliability
If compression is applied to all data links, then network congestion is reduced, but system complexity and implementation difficulty increase
Solution Approach 1:
Instead of applying compression to all data links uniformly, the system applies compression partially - only to those data links where the storage device has sufficient available CPU resources. The networking device monitors resource availability and selectively enables compression on a per-link basis, avoiding the complexity of universal compression while still achieving congestion relief where possible
Data Source
AI summary
The present disclosure relates to a method for data communication between nodes and storage devices via a networking device of a network, wherein data access to a storage device of the network by a node is performed in accordance with a data access mode being a normal data access mode. The networking device is configured to share available resources of the networking device between the data links. The method comprises: determining that a network congestion may occur in the network. At least one data link defined by a node and a storage device may be identified. The data access mode may be switched for the identified data link from the normal data access mode to a compressed data access mode. The compressed data access mode enables exchange of compressed data via the identified data link, thereby increasing available resources of the networking device.