Step-by-step back pressure mechanism for RDMA storage network and implementation method of step-by-step back pressure mechanism

Abstract

The invention discloses a step-by-step back pressure mechanism for an RDMA (Remote Direct Memory Access) storage network and an implementation method thereof, and relates to the technical field of computer network communication, and the step-by-step back pressure mechanism comprises a hardware architecture which comprises a back pressure signal generator for generating a back pressure signal in real time according to a network congestion condition and a signal transmission line; the receiving module is used for receiving the back pressure signal, the network congestion condition and the data flow; and the processing module is used for processing the back pressure signal and calculating the intensity and the propagation range of the back pressure signal based on the network congestion condition and the data flow. By designing a special hardware architecture, optimizing a storage node structure and developing an efficient algorithm, quick response and accurate control of network congestion can be realized. A back voltage signal generator and a signal transmission line ensure timely generation and accurate propagation of back voltage signals, and a back voltage signal receiving module and a processing unit of the storage node improve the response speed.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer network communication, and in particular relates to a step-by-step back pressure mechanism for an RDMA storage network and an implementation method thereof. Background Art

[0002] Remote Direct Memory Access (RDMA) technology significantly improves data transmission efficiency by allowing network adapters to directly access memory without CPU intervention. Although RDMA technology has obvious advantages, when faced with a surge in data traffic or network congestion, traditional RDMA storage networks often suffer from data loss and increased transmission delays due to the lack of an effective back pressure mechanism.

[0003] Existing back pressure methods are inefficient and complex to implement, and are difficult to meet the needs of modern high-performance storage networks. Challenges faced by RDMA networks include insufficient response to burst traffic and limitations in traffic management;

[0004] Existing flow control algorithms and back pressure solutions often limit the improvement of network performance due to slow response speed, lack of intelligent scheduling, hardware support or complexity of software implementation. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a step-by-step back pressure mechanism for an RDMA storage network and an implementation method thereof, which can achieve rapid response and precise control of network congestion by designing a special hardware architecture, optimizing the storage node structure, and developing an efficient algorithm. The back pressure signal generator and the signal transmission line ensure the timely generation and accurate propagation of the back pressure signal, and the back pressure signal receiving module and processing unit of the storage node improve the response speed.

[0006] To achieve the above object, the present invention proposes a step-by-step back pressure mechanism for an RDMA storage network, which is characterized by comprising:

[0007] A hardware architecture, including a back pressure signal generator for generating a back pressure signal in real time according to network congestion conditions, and a signal transmission line;

[0008] A receiving module, used for receiving back pressure signals, network congestion conditions and data traffic;

[0009] The processing module is used to process the back pressure signal and calculate the strength and propagation range of the back pressure signal based on the network congestion situation and data flow.

[0010] Preferably, the strength calculation expression of the back pressure signal is: Among them, YC is the back pressure signal strength, J0 is the basic strength coefficient, D0 is the current network congestion level, and DMAX T is the maximum tolerable congestion level and the adjustment index.

[0011] Preferably, the current network congestion level is calculated by calculating the data flow, buffer occupancy and transmission delay in the storage network. The specific calculation method is: C = W F ×F+W O ×O+W D ×D, where F is the data flow, O is the buffer occupancy, D is the transmission delay, and W F , W O and W D are the weights of data flow, buffer occupancy and transmission delay respectively.

[0012] Preferably, the adjustment index is used to control the speed at which the back pressure signal strength changes with the degree of network congestion, and the value range is a value greater than 0 and less than 1.

[0013] The method for implementing the step-by-step back pressure for the RDMA storage network is based on the step-by-step back pressure mechanism for the RDMA storage network, and includes:

[0014] Real-time monitoring of network status, including data flow, buffer occupancy, and transmission delay, and comprehensive assessment of network congestion based on data flow, buffer occupancy, and transmission delay;

[0015] The congestion thresholds are preset in stages, namely, congestion threshold one, congestion threshold two and congestion threshold three. The network congestion degree is compared with the congestion threshold one, congestion threshold two and congestion threshold three respectively to determine whether to generate an adjustment instruction.

[0016] Preferably, a priority processing method for the back pressure signal is also included, specifically:

[0017] Obtain back pressure signals of n storage nodes, where n is an integer greater than 1;

[0018] The priorities of the n storage nodes are obtained, and the processing order of the back pressure signal is determined based on the priorities of the storage nodes.

[0019] Preferably, the priority determination method is:

[0020] Obtain the importance of the storage node, the urgency of data transmission, and set the importance threshold and urgency threshold;

[0021] Compare the importance and urgency with the importance threshold and urgency threshold respectively;

[0022] If the importance is less than or equal to the importance threshold, and the urgency is less than or equal to the urgency threshold, the priority is determined to be low;

[0023] If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be medium;

[0024] If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be high.

[0025] Preferably, the back pressure signal adjustment method is triggered to have different degrees:

[0026] When the network congestion is light, the data transmission rate is only slightly reduced;

[0027] When the network is heavily congested, the data transmission rate is significantly reduced;

[0028] When the network is severely congested, data transmission is suspended and the transmission is resumed after the network congestion is alleviated.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. By designing a specialized hardware architecture, optimizing the storage node structure, and developing efficient algorithms, it is possible to achieve rapid response and precise control of network congestion. The back pressure signal generator and signal transmission line ensure the timely generation and accurate propagation of the back pressure signal, while the back pressure signal receiving module and processing unit of the storage node improve the response speed;

[0031] 2. The strength calculation formula of the back pressure signal takes into account multiple factors and can dynamically adjust the back pressure signal strength according to the actual network conditions, thus improving the adaptability and effectiveness of the back pressure mechanism;

[0032] 3. The introduction of multiple mechanisms such as dynamic adjustment mechanism, priority processing, timeout processing, error processing, encryption processing, authentication processing and statistical analysis further improves the performance and reliability of the back pressure mechanism. The dynamic adjustment mechanism can flexibly adjust the back pressure measures according to the degree of network congestion; priority processing ensures that important data transmission is not affected; timeout and error processing improve the stability of signal transmission; encryption and authentication processing ensure the security and authenticity of the signal; statistical analysis helps to optimize the back pressure mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The figure is a flow chart of the mechanism of the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Example

[0036] like Figure 1 As shown in FIG. 1 , the step-by-step back pressure mechanism for the RDMA storage network includes:

[0037] The hardware architecture includes a back pressure signal generator for generating back pressure signals in real time according to network congestion conditions, and a signal transmission line. The back pressure signal generator has a highly sensitive monitoring mechanism that can continuously monitor various key parameters in the storage network in real time, such as data flow, buffer occupancy, transmission delay, etc. Once signs of network congestion are detected, the back pressure signal generator will start quickly and generate a back pressure signal of corresponding strength according to specific algorithms and rules. The signal transmission line is responsible for efficiently and stably transmitting the generated back pressure signal to the receiving module, ensuring that the back pressure signal can be transmitted to the entire storage network system in a timely and accurate manner to achieve effective control of network congestion;

[0038] The receiving module is used to receive back pressure signals, network congestion conditions and data traffic, and can receive information from different channels at the same time. For back pressure signals, the receiving module can quickly identify their strength and source, and pass them to the processing module for further analysis and processing. For network congestion conditions, the receiving module communicates with the back pressure signal generator in the hardware architecture and other monitoring devices in the storage network to obtain the latest status of network congestion in real time, including key parameters such as data traffic, buffer occupancy, and transmission delay. For data traffic, the receiving module can accurately monitor and record data transmission in the storage network, providing accurate data support for the processing module;

[0039] The processing module is used to process the back pressure signal and calculate the strength and propagation range of the back pressure signal based on the network congestion and data flow. The processing module first analyzes the back pressure signal transmitted by the receiving module to determine its initial strength and source. Then, the processing module combines the network congestion and data flow information provided by the receiving module, and adopts a specific calculation method, such as weighted average, to calculate the strength of the back pressure signal by assigning different weights to each parameter according to the degree of influence on the network congestion and then summing them up. At the same time, the processing module will also calculate the propagation range of the back pressure signal based on factors such as the topological structure of the storage network, the node distribution, and the propagation characteristics of the back pressure signal, to ensure that the back pressure signal can be effectively transmitted to the node that needs to be controlled, thereby achieving precise control of the storage network congestion.

[0040] Exemplary, back pressure signal generator design, sensor integration: integrated data flow sensor, capable of real-time monitoring of data inflow and outflow speed in the network. A high-speed counter or flow monitoring chip can be used to accurately measure the size of data flow; install a buffer occupancy sensor to determine the occupancy by monitoring the storage status of the buffer. Capacitive or resistive sensors can be used to generate corresponding electrical signals according to the filling level of the buffer; configure a transmission delay sensor, and use clock synchronization technology and signal round-trip time measurement methods to accurately measure the transmission delay of data in the network. For example, a high-precision timer and signal reflection technology can be used to determine the delay time; signal processing unit: a high-performance microprocessor or digital signal processor (DSP) is used as the core processing unit, which is responsible for receiving sensor data and performing real-time analysis. The processor should have sufficient computing power and processing speed to meet the needs of rapid response to network congestion; design special algorithms and logic circuits to generate back pressure signals based on sensor data. For example, according to the threshold settings of data flow, buffer occupancy and transmission delay, the generation of back pressure signals can be triggered when a certain threshold is exceeded; add a signal conditioning circuit to filter, amplify and shape the generated back pressure signal to ensure the stability and reliability of the signal;

[0041] Communication interface: Provides a variety of communication interfaces to connect with signal transmission lines and other devices. Common interfaces include Ethernet interface, USB interface, serial port, etc. You can choose the appropriate interface type according to actual needs; supports standard communication protocols such as TCP / IP, UDP, etc., to ensure that back pressure signals can be effectively transmitted and interacted between different devices;

[0042] Signal transmission line design, physical medium selection: Select the appropriate physical medium as the signal transmission line based on factors such as transmission distance, bandwidth requirements and anti-interference ability. Common choices include twisted pair, optical fiber and coaxial cable; for short-distance transmission, twisted pair can be used, which has the advantages of low cost and easy installation. For long-distance transmission or high bandwidth requirements, optical fiber is a better choice, with the characteristics of fast transmission speed and strong anti-interference ability. Signal transmission method: Use digital signal transmission to ensure the accuracy and reliability of the back pressure signal. Pulse code modulation (PCM) or other digital coding methods can be used to convert the back pressure signal into a digital signal for transmission.

[0043] It is worth mentioning that differential signal transmission technology should be considered to improve the signal's anti-interference ability. For example, differential twisted pair or differential optical fiber can be used to reduce the impact of external interference by transmitting two complementary signals. Signal relay and amplification: Set up a signal repeater or amplifier in the signal transmission line to enhance the signal strength and transmission distance. The repeater can regenerate and amplify the signal to ensure that the signal does not attenuate too much during long-distance transmission. According to actual needs, the position of the repeater or amplifier should be reasonably arranged to ensure that the signal can be stably transmitted to the receiving module.

[0044] Reliability design and redundancy design of hardware architecture: In order to improve the reliability of hardware architecture, redundant design can be adopted. For example, multiple back pressure signal generators and signal transmission lines can be set up. When one of them fails, other devices can continue to work to ensure the stability of the system; redundant power supply is adopted to ensure that the system can still operate normally when the main power fails. Backup batteries or uninterruptible power supply (UPS) and other equipment can be used to provide temporary power support.

[0045] Fault detection and alarm: Design a fault detection mechanism to monitor the operating status of each device and component in the hardware architecture in real time. Sensors, monitoring circuits, or software algorithms can be used to detect equipment failures, signal loss, or abnormal conditions; when a fault is detected, an alarm signal is issued in a timely manner to notify the system administrator to handle it. Alarm methods can include sound and light alarms, email notifications, or text message notifications, so that administrators can take timely measures.

[0046] Furthermore, the strength calculation expression of the back pressure signal is: Among them, YC is the back pressure signal strength, and this strength value determines the degree of adjustment of data traffic by each node in the network, thereby effectively alleviating network congestion; J0 is the basic strength coefficient, which represents the initial strength level of the back pressure signal in the absence of network congestion. This coefficient is usually set according to the characteristics and design requirements of the network, and a suitable value can be determined through experiments and simulations. The size of the basic strength coefficient will affect the propagation speed and effect of the back pressure signal in the network. Too large or too small may lead to unstable network performance; D0 is the current network congestion level, which is determined by monitoring and analyzing multiple key parameters such as data traffic, buffer occupancy, and transmission delay in the storage network. The specific calculation method can adopt a weighted average method, which assigns different weights to each parameter according to the degree of influence on network congestion and then sums them up. Accurate assessment of the current network congestion level is crucial to calculating the back pressure signal strength, because it directly reflects the actual operating status of the network; D MAXThe maximum tolerable congestion level indicates the maximum congestion level that the network can withstand without serious performance degradation. This value is usually determined based on factors such as the network's design capacity, service quality requirements, and user needs. When the network congestion level exceeds the maximum tolerable congestion level, network performance may drop sharply, or even cause service interruption; T is the adjustment index, which is used to adjust the rate at which the back pressure signal strength changes with the network congestion level. The size of the adjustment index can be adjusted according to the dynamic characteristics of the network and the response speed requirements for congestion control. A larger adjustment index will make the back pressure signal strength more sensitive to changes in the network congestion level, thereby taking congestion control measures more quickly, but it may also lead to overreaction; a smaller adjustment index will make the back pressure signal strength change relatively slowly, and it may take longer to cope with network congestion. The back pressure signal strength calculated by this expression can be adjusted in real time according to the network congestion situation, thereby effectively controlling the data traffic in the network and ensuring the stable operation of the network. In practical applications, it can also be combined with other congestion control mechanisms, such as traffic shaping and queue management, to further improve the performance and reliability of the network.

[0047] Specifically, the current network congestion level is calculated by calculating the data flow, buffer occupancy and transmission delay in the storage network. The specific calculation method is: C = W F ×F+W O ×O+W D ×D, where F is the data flow, O is the buffer occupancy, D is the transmission delay, and W F , W O and W D are the weights of data flow, buffer occupancy and transmission delay respectively; illustratively, data flow F reflects the activity of data transmission in the network. A higher data flow may mean a heavier network load, which is prone to cause congestion. Data flow can be measured in real time by network monitoring equipment or software. The unit is usually bits per second (bps) or bytes per second (Bps). Different types of network applications and services will generate different data flow patterns. For example, applications such as video streaming, file downloading and online games may generate larger data flows;

[0048] The buffer occupancy rate O indicates the usage of the buffer in the storage network. Buffers are usually used to temporarily store data to smooth fluctuations in data transmission. When the buffer occupancy rate is high, it means that there is a large backlog of data in the network, which may cause increased transmission delays and increased packet loss rates. The buffer occupancy rate can be determined by monitoring the storage status of the buffer, usually expressed as a percentage;

[0049] Transmission delay D refers to the time delay experienced by data from the sender to the receiver. Increased transmission delay may be caused by network congestion, improper routing or device performance degradation. Transmission delay can be determined by measuring the round-trip time (RTT) of the data packet, usually in milliseconds (ms).

[0050] Furthermore, the adjustment index is used to control the speed at which the back pressure signal strength changes with the degree of network congestion, and the value range is a value greater than 0 and less than 1.

[0051] The method for implementing the step-by-step back pressure for the RDMA storage network is based on the step-by-step back pressure mechanism for the RDMA storage network, and includes:

[0052] Monitor the network status in real time, including data flow, buffer occupancy and transmission delay, and comprehensively evaluate the degree of network congestion based on data flow, buffer occupancy and transmission delay. For example, use professional network monitoring equipment or software to monitor the data flow in the RDMA storage network in real time. These devices can accurately measure the size of data flow by analyzing parameters such as the number, size and transmission rate of network data packets. For example, if the data flow suddenly increases, it may mean that a large amount of data is being transmitted, which may cause network congestion;

[0053] The congestion thresholds are preset in different levels, namely congestion threshold 1, congestion threshold 2 and congestion threshold 3. The network congestion level is compared with congestion threshold 1, congestion threshold 2 and congestion threshold 3 to determine whether to generate an adjustment instruction. For example, the determination of the congestion threshold needs to consider the performance requirements and application requirements of the network. Generally speaking, the congestion threshold should be set high enough to avoid unnecessary back pressure operations, but not too high to avoid affecting the performance of the network. The congestion threshold is graded as follows:

[0054] The congestion threshold is divided into different levels, such as congestion threshold one, congestion threshold two and congestion threshold three. Different levels of congestion thresholds correspond to different back pressure strengths; the congestion threshold of this embodiment is the lowest level of congestion threshold. When the network congestion level reaches congestion threshold one, only weaker back pressure measures need to be taken, such as reducing the data transmission rate or adjusting the routing selection. Congestion threshold two and congestion threshold three are higher levels of congestion thresholds. When the network congestion level reaches these thresholds, stronger back pressure measures need to be taken, such as suspending data transmission or rerouting data.

[0055] Furthermore, a priority processing method for back pressure signals is also included, specifically:

[0056] Obtain back pressure signals of n storage nodes, where n is an integer greater than 1;

[0057] The priorities of the n storage nodes are obtained, and the processing order of the back pressure signal is determined based on the priorities of the storage nodes.

[0058] Specifically, the priority determination method is:

[0059] Obtain the importance of the storage node, the urgency of data transmission, and set the importance threshold and urgency threshold;

[0060] Compare the importance and urgency with the importance threshold and urgency threshold respectively;

[0061] If the importance is less than or equal to the importance threshold, and the urgency is less than or equal to the urgency threshold, the priority is determined to be low;

[0062] If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be medium;

[0063] If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be high.

[0064] It is worth mentioning that the adjustment method of triggering different degrees of back pressure signal is:

[0065] When the network congestion is light, the data transmission rate is only slightly reduced;

[0066] When the network is heavily congested, the data transmission rate is significantly reduced;

[0067] When the network is severely congested, data transmission is suspended and the transmission is resumed after the network congestion is alleviated.

[0068] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0069] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0070] The above preset parameters or preset thresholds are all set by technicians in this field according to actual conditions or obtained through large-scale data simulation.

[0071] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A step-by-step back pressure mechanism for RDMA storage networks, characterized in that: include: A hardware architecture, including a back pressure signal generator for generating a back pressure signal in real time according to network congestion conditions, and a signal transmission line; A receiving module, used for receiving back pressure signals, network congestion conditions and data traffic; The processing module is used to process the back pressure signal and calculate the strength and propagation range of the back pressure signal based on the network congestion situation and data flow.

2. The step-by-step back pressure mechanism for RDMA storage network according to claim 1, characterized in that: The strength calculation expression of the back pressure signal is: Among them, YC is the back pressure signal strength, J0 is the basic strength coefficient, D0 is the current network congestion level, and D MAX T is the maximum tolerable congestion level and the adjustment index.

3. The step-by-step back pressure mechanism for RDMA storage network according to claim 2, characterized in that: The current network congestion level is calculated by the data flow, buffer occupancy and transmission delay in the storage network. The specific calculation method is: C = W F ×F+W O ×O+W D ×D, where F is the data flow, O is the buffer occupancy, D is the transmission delay, and W F , W O and W D are the weights of data flow, buffer occupancy and transmission delay respectively.

4. The step-by-step back pressure mechanism for RDMA storage network according to claim 3, characterized in that: The adjustment index is used to control the speed at which the back pressure signal strength changes with the degree of network congestion, and the value range is a value greater than 0 and less than 1.

5. A method for implementing step-by-step back pressure for an RDMA storage network, based on the step-by-step back pressure mechanism for an RDMA storage network according to any one of claims 1 to 4, characterized in that: include: Real-time monitoring of network status, including data flow, buffer occupancy, and transmission delay, and comprehensive assessment of network congestion based on data flow, buffer occupancy, and transmission delay; The congestion thresholds are preset in stages, namely, congestion threshold one, congestion threshold two and congestion threshold three. The network congestion degree is compared with the congestion threshold one, congestion threshold two and congestion threshold three respectively to determine whether to generate an adjustment instruction.

6. The method for implementing step-by-step back pressure for an RDMA storage network according to claim 5, characterized in that: It also includes a priority processing method for back pressure signals, specifically: Obtain back pressure signals of n storage nodes, where n is an integer greater than 1; The priorities of the n storage nodes are obtained, and the processing order of the back pressure signal is determined based on the priorities of the storage nodes.

7. The method for implementing step-by-step back pressure for an RDMA storage network according to claim 6, characterized in that: The priority is determined as follows: Obtain the importance of the storage node, the urgency of data transmission, and set the importance threshold and urgency threshold; Compare the importance and urgency with the importance threshold and urgency threshold respectively; If the importance is less than or equal to the importance threshold, and the urgency is less than or equal to the urgency threshold, the priority is determined to be low; If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be medium; If the importance is greater than the importance threshold and the urgency is less than or equal to the urgency threshold, the priority is determined to be high.

8. The method for implementing step-by-step back pressure for an RDMA storage network according to claim 7, characterized in that: Trigger different degrees of back pressure signal adjustment method: When the network congestion is light, the data transmission rate is only slightly reduced; When the network is heavily congested, the data transmission rate is significantly reduced; When the network is severely congested, data transmission is suspended and the transmission is resumed after the network congestion is alleviated.

Images