RDMA Packet Compression for Higher Bandwidth Utilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current RDMA technologies face challenges in maximizing network bandwidth utilization and transmission efficiency due to the high volume of data transmitted over networks, leading to suboptimal performance in data centers.
Innovation Solution
Implementing a method where RDMA devices compress raw data using a preset compression method, encapsulate the compressed data into packets with a method tag, and transmit these packets, allowing the receiving end to decompress the data using the corresponding algorithm, thereby reducing the data volume and improving transmission efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is transmitted without compression in RDMA networks, then data transmission simplicity is maintained, but network bandwidth utilization and transmission efficiency deteriorate due to high data volume
Solution Approach 1:
The patent applies preliminary action by performing data compression before transmission. The sending-end RDMA device compresses the data using a compression algorithm and encapsulates it into a data packet with a method tag indicating compression was applied. This pre-compression reduces the data volume before it enters the network, improving bandwidth utilization and transmission efficiency without affecting the receiving end's processing complexity.
2Productivity
If data compression is applied at the sending end, then data volume is reduced and transmission efficiency is improved, but device complexity increases due to additional compression and decompression operations
Solution Approach 1:
The patent implements self-service by enabling the RDMA devices themselves to perform compression and decompression operations autonomously. The sending-end RDMA device automatically compresses data and sets the method tag, while the receiving-end RDMA device automatically detects the method tag and performs corresponding decompression. This eliminates the need for external compression software or manual intervention, reducing overall system complexity while maintaining high transmission efficiency.
3Reliability
If compression method tags are added to data packets, then receiving end can accurately decompress data, but packet structure complexity increases
Solution Approach 1:
The patent applies local quality by adding the method tag only to the packet header rather than modifying the entire packet structure. The method tag is a compact field that indicates whether compression was applied and what type of compression was used. This localized addition provides the receiving end with sufficient information to perform accurate decompression while minimizing the increase in overall packet structure complexity.
Data Source
AI summary
A Remote Direct Memory Access (RDMA) data sending method is disclosed. The method is applicable to a sending end, with a data-transmission RDMA device disposed thereon. The method includes: the data-transmission RDMA device acquiring raw data; the data-transmission RDMA device compressing the raw data by using a preset compression method to obtain compressed data; and the data-transmission RDMA device encapsulating the compressed data into a data packet, and transmitting the data packet to a receiving end. The data packet may include a method tag corresponding to the preset compression method. In this method, the compression and transmission of the raw data are conducted by the data-transmission RDMA device on the hardware level.