Massive distributed storage system

Abstract

The application discloses a large-capacity distributed storage system and belongs to the field of storage systems, comprising a client module, a dynamic load balancing module, a node cluster and a monitoring module. The client module is used for receiving the request of a user; the monitoring module is used for collecting and analyzing the real-time node performance data, reliable performance data and expansion performance data of the system to generate risk signals and maintenance signals to determine the real-time state of the system; the dynamic load balancing module is used for dynamically distributing the request to multiple nodes in the node cluster, processing the risk problems of the system in time when the risk signals are received, analyzing and predicting the current load data when the safety signals are received, adjusting the load distribution strategy in advance according to the prediction result, reasonably distributing the request and avoiding the excessive load of the system to realize the load balancing of the system and ensure the efficient and stable operation of the system; the node cluster is composed of multiple nodes and can be horizontally expanded to meet the actual storage demand.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of storage systems, and particularly relates to a large-capacity distributed storage system. BACKGROUND

[0002] The storage system refers to a combination of hardware and software for storing and managing data. It can be a single device or a cluster or network composed of multiple devices; load balancing technology is a technology for distributing and managing network traffic, which is divided into two forms of hardware load balancing and software load balancing.

[0003] At present, when storing data, the node performance data, reliable performance data and expansion performance data of the storage system cannot be monitored and collected in real time, and then risk signals and maintenance signals are generated through in-depth analysis to determine the real-time state of the system and analyze and predict the current load data, and according to the prediction result, the load distribution strategy is adjusted in advance.

[0004] In view of the above technical defects, a solution is proposed. SUMMARY

[0005] The purpose of the present application is to provide a large-capacity distributed storage system to solve the technical problem that the node performance data, reliable performance data and expansion performance data of the storage system cannot be monitored and collected in real time, and then risk signals and maintenance signals are generated through in-depth analysis to determine the real-time state of the system and analyze and predict the current load data, and according to the prediction result, the load distribution strategy is adjusted in advance.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0007] The large-capacity distributed storage system comprises a client module, a dynamic load balancing module, a node cluster and a monitoring module; wherein the client module and the dynamic load balancing module are in bidirectional communication connection; the dynamic load balancing module and the node cluster are in unidirectional communication connection; the monitoring module and the dynamic load balancing module are in bidirectional communication connection; the monitoring module and the node cluster are in bidirectional communication connection; the node cluster is composed of multiple horizontally scalable nodes; the dynamic load balancing module comprises an evaluation module, an adjustment module and a distribution module.

[0008] The client module is used for receiving the request of the user.

[0009] The dynamic load balancing module is used for receiving the request sent by the client module, and obtaining the load condition of the current system from the monitoring module, and dynamically distributing the request to multiple nodes in the node cluster.

[0010] Node cluster: used for dispersing data storage on nodes, using distributed storage technology, increasing storage capacity and performance, node cluster is composed of multiple nodes, which can be horizontally expanded to meet the actual storage needs of the system.

[0011] Monitoring module: used for real-time monitoring and collecting node performance data, reliability performance data and expansion performance data of the system, and conducting in-depth analysis to generate risk signals and maintenance signals to determine the real-time state of the system.

[0012] Preferably, the functions of the evaluation module, the adjustment module and the distribution module in the dynamic load balancing module are as follows:

[0013] Evaluation module: used for analyzing and determining the received risk signals and maintenance signals, generating various signals and sending them to the adjustment module and the distribution module.

[0014] Adjustment module: used for receiving abnormal alarm signals and abnormal prompt signals sent by the evaluation module, and formulating load scheduling strategies according to the real-time state of the system and sending them to the distribution module.

[0015] Distribution module: used for receiving safety signals sent by the evaluation module and load scheduling strategies sent by the adjustment module, analyzing and predicting the current load data when receiving the safety signals, adjusting the load distribution strategy in advance, and reasonably distributing the requests according to the prediction results.

[0016] Preferably, the workflow of the monitoring module is as follows:

[0017] S1, collecting node performance data, reliability performance data and expansion performance data of the large-capacity distributed storage system; the node performance data includes CPU utilization, memory usage, disk I / O utilization; the reliability performance data includes mean time between failures, mean time to repair; the expansion performance data includes performance loss ratio, throughput growth ratio;

[0018] S2, when any of the following conditions occurs: CPU utilization is continuously greater than 70% ~ 80%, memory usage is greater than 80% ~ 90%, disk I / O utilization is greater than 70%, and request response time is continuously greater than the set 500 millisecond threshold, it is determined that the node performance data has a level I node risk, and a level I node risk signal is generated; when CPU utilization is less than or equal to 70%, memory usage is less than or equal to 80%, disk I / O utilization is less than or equal to 70%, and request response time is less than or equal to the set 500 millisecond threshold, it is determined that the node performance data has no risk, and a node maintenance signal is generated; when the mean time between failures is less than the set threshold or the mean time to repair is greater than the set threshold When the average failure-free time is greater than or equal to the set threshold value and the average repair time is less than or equal to the set threshold value, it is determined that the reliability performance data does not have a risk, and a reliable maintenance signal is generated; when the performance loss ratio is high and the throughput growth ratio is low, it is determined that the expansion performance data has a first-level expansion risk, and a first-level expansion risk signal is generated; when the performance loss ratio is high and the throughput growth ratio is high, it is determined that the expansion performance data has a second-level expansion risk, and a second-level expansion risk signal is generated; when the performance loss ratio is low and the throughput growth ratio is high, it is determined that the expansion performance data does not have a risk, and an expansion maintenance signal is generated; and each type of signal generated above is sent to the dynamic load balancing module.

[0019] Preferably, the workflow of the evaluation module is as follows: when any one of the first-level node risk signal, the first-level reliability risk signal and the first-level expansion risk signal is received, it is determined that the current node load of the system is too high, an abnormal alarm signal is generated, and the abnormal alarm signal is sent to the adjustment module; when the second-level expansion risk signal is received, it is determined that the node load is high, an abnormal prompt signal is generated, and the abnormal prompt signal is sent to the adjustment module; when the node maintenance signal, the reliable maintenance signal and the expansion maintenance signal are received at the same time, it is determined that the node load is normal, a safety signal is generated, and the safety signal is sent to the distribution module.

[0020] Preferably, the scheduling strategy formulated by the adjustment module is as follows: when the abnormal alarm signal caused by the first-level node risk signal is received, the node load with the number i marked as having a risk is adjusted and migrated to other normal nodes; when the node load is high, the weight of the node is reduced, and the proportion of requests distributed to the node is reduced; when the abnormal alarm signal caused by the first-level reliability risk signal is received, the system data is regularly backed up; when the abnormal alarm signal caused by the first-level expansion risk signal and the abnormal prompt signal caused by the second-level expansion risk signal are received, the node cluster is horizontally expanded to meet the actual storage demand.

[0021] Preferably, the adjustment load distribution strategy workflow of the distribution module is as follows:

[0022] S21, constructing a load redundancy value binary search tree: collecting load redundancy values, and the process of obtaining the load redundancy values is specifically as follows: the current load value of a node is brought into the formula wherein f is the current load value of the node, q is the redundancy requirement value of the node, is a correction factor, and The first load redundancy value j is taken as a root node, and the second load redundancy value is sequentially inserted into the binary search tree, and when each node is inserted, the size relationship between the node value and the current node is compared, if it is smaller than the current node, it is continuously inserted in the left subtree of the current node, if it is greater than the current node, it is continuously inserted in the right subtree of the current node, and the process is repeated until all nodes are inserted into the binary search tree;

[0023] S22, in-order traversal of the load redundancy value binary search tree: in-order traversal is performed on the generated binary search tree, and the traversal result is an ascending order list of sorted load redundancy values;

[0024] S23, request distribution: according to the sorted load redundancy value sorting table, each node of the node cluster is assigned a weight value ki, ki is a number greater than 0, an ascending order list of node weight values is generated, the weight of the low weight node is increased, the proportion of the request distributed to the node is improved, and the load is improved; when a new request arrives at the dynamic load balancing module, the distribution module checks the current connection number and weight value of each node in the node cluster, and selects the node with the smallest connection number and weight value to process the request.

[0025] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present application are:

[0026] 1. Based on the performance of the storage system, the present application collects and monitors the node performance data, reliability performance data and expansion performance data in real time, avoids wasting time and resources on unnecessary data collection, can timely discover abnormal conditions and potential problems in the system, and improves work efficiency;

[0027] 2. The present application determines the load level of the system node by analyzing the three types of data in depth and generating risk signals and maintenance signals, generates corresponding abnormal alarm signals, abnormal prompt signals and safety signals, formulates a reasonable load scheduling strategy through the abnormal alarm signals and abnormal prompt signals, ensures the availability and durability of data, and analyzes and predicts the current load data through the safety signals, adjusts the load distribution strategy in advance, reasonably allocates requests according to the prediction results, avoids excessive load of the system, and ensures efficient and stable operation of the system. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0029] Figure 1 The overall flowchart of the present application is shown;

[0030] Figure 2 The flowchart of the dynamic load balancing module of the present application is shown. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application. EMBODIMENT

[0032] Referring to Figure 1 The large-capacity distributed storage system of the present embodiment includes a client module, a dynamic load balancing module, a node cluster and a monitoring module. The client module and the dynamic load balancing module are in bidirectional communication connection; the dynamic load balancing module and the node cluster are in unidirectional communication connection; the monitoring module and the dynamic load balancing module are in bidirectional communication connection; the monitoring module and the node cluster are in bidirectional communication connection; the node cluster is composed of multiple nodes and can be horizontally expanded to meet the actual storage requirements; the dynamic load balancing module includes an evaluation module, an adjustment module and a distribution module.

[0033] The client module is used to receive the request of a user, send the request to the node cluster, obtain the metadata information of a file, and then establish a connection with a storage node according to the metadata information and perform read / write operation on data.

[0034] The dynamic load balancing module is used to receive the request sent by the client module, obtain the load condition of the current system from the monitoring module, and dynamically distribute the request to multiple nodes in the node cluster to achieve system load balancing.

[0035] The node cluster is used to store data on nodes, uses distributed storage technology to increase storage capacity and performance, and includes metadata nodes and storage nodes. The node cluster is composed of multiple nodes and can be horizontally expanded to meet the actual storage requirements of the system.

[0036] The monitoring module is used to perform real-time monitoring and collection on three kinds of data, i.e., node performance data, reliability performance data and expansion performance data, and perform in-depth analysis to generate risk signals and maintenance signals to determine the real-time state of the system. The working process of the monitoring module is as follows:

[0037] S1, collect node performance data, reliability performance data and expansion performance data of the large-capacity distributed storage system; the node performance data includes CPU utilization, memory usage and disk I / O utilization; the reliability performance data includes mean time between failures and mean time to repair; the expansion performance data includes performance loss ratio and throughput growth ratio;

[0038] S2, when any of the following conditions occurs: CPU utilization continuously greater than 70% ~ 80%, memory usage greater than 80% ~ 90%, disk I / O utilization greater than 70%, and request response time continuously greater than the set 500 ms threshold, determine that the node performance data has a level I node risk, generate a level I node risk signal, and send it to the dynamic load balancing module; when the CPU utilization is less than or equal to 70%, the memory usage is less than or equal to 80%, the disk I / O utilization is less than or equal to 70%, and the request response time is less than or equal to the set 500 ms threshold, determine that the node performance data has no risk, generate a node maintenance signal, and send it to the dynamic load balancing module;

[0039] When the mean time between failures is less than the set threshold or the mean time to repair is greater than the set threshold , determine that the reliability performance data has a level I reliability risk, generate a level I reliability risk signal; when the mean time between failures is greater than or equal to the set threshold and the mean time to repair is less than or equal to the set threshold, determine that the reliability performance data has no risk, generate a reliability maintenance signal, and send it to the dynamic load balancing module;

[0040] When the performance loss ratio is greater than the expected threshold Q1, it means that the performance of the system after expansion is affected to a certain extent; when the throughput growth ratio is greater than the expected threshold Q2, it means that the system can effectively handle more tasks and requests after expansion; when the performance loss ratio is high and the throughput growth ratio is low, determine that the expansion performance data has a level I expansion risk, generate a level I expansion risk signal, and send it to the dynamic load balancing module; when the performance loss ratio is high and the throughput growth ratio is high, determine that the expansion performance data has a level II expansion risk, generate a level II expansion risk signal, and send it to the dynamic load balancing module; when the performance loss ratio is low and the throughput growth ratio is high, determine that the expansion performance data has no risk, generate an expansion maintenance signal, and send it to the dynamic load balancing module.

[0041] CPU utilization: CPU is one of the core components of the server to process requests. By monitoring the CPU utilization, the load of the server can be understood. High CPU utilization indicates that the server is performing a large number of computing tasks and may require more processing power.

[0042] Memory usage: Memory is a critical resource for servers to store data and programs. By monitoring the memory usage, we can understand the current memory pressure of the server. High memory usage may cause requests to be delayed or rejected, so we can allocate requests based on memory usage.

[0043] Disk I / O utilization: Disk I / O is an important indicator for servers to read and write data. By monitoring the disk I / O utilization, we can understand the access pressure of the server to the disk. High disk I / O utilization may indicate that the server is frequently reading and writing large amounts of data, so we can perform load balancing scheduling based on disk I / O utilization.

[0044] Request response time: Monitoring the response time of the server to the request can understand the load situation of the server. Longer response time may indicate that the server's ability to handle requests is limited, and we can consider allocating requests to servers with shorter response times.

[0045] Mean Time Between Failures (MTBF): It refers to the average time that a system or device runs without failure within a certain period of time. It is calculated by dividing the total working time before all failures by the number of failures. The longer the MTBF, the higher the reliability of the system, and the lower the probability of failure.

[0046] Mean Time to Repair (MTTR): It refers to the average time required to repair a system or device after it fails. It is calculated by dividing the total time required to repair all failures by the number of failures. The shorter the MTTR, the higher the recoverability of the system, meaning that the system can recover to normal operation more quickly.

[0047] Performance loss ratio: It is used to measure the degree of performance loss of the system during expansion. It represents the proportion of performance reduction compared with the original system after the system is expanded. Generally, we hope that the performance loss ratio is lower, that is, we want to reduce the impact of expansion on system performance as much as possible.

[0048] Throughput growth ratio: It is used to measure the growth ratio of the system's throughput during expansion. It represents the proportion of throughput growth compared with the original system after the system is expanded. Higher throughput growth ratio means that the system can effectively utilize the expanded resources to achieve higher throughput.

[0049] The monitoring module in the embodiment of the application monitors and collects node performance data, reliability performance data and expansion performance data of the system based on the performance of the storage system, so that time and resources are not wasted on unnecessary data collection, the system can be operated more quickly, potential problems of the storage system can be found in time and efficient and stable operation of the system is ensured. The three types of data are analyzed in depth and risk signals and maintenance signals are generated to determine the real-time state of the system, so that the dynamic load balancing module in embodiment 2 can take corresponding measures according to the current system state. Embodiment

[0050] Referring to Figure 2 The dynamic load balancing module receives the first-level node risk signal, the first-level reliability risk signal, the first-level expansion risk signal, the second-level expansion risk signal, the node maintenance signal, the reliability maintenance signal and the expansion maintenance signal sent by the monitoring module, and obtains the state of the current system. After processing the request, the dynamic load balancing module updates the connection number of each node in the node cluster and returns the request result to the client module.

[0051] The evaluation module is used to analyze and determine the received risk signals and maintenance signals, generate various signals and send them to the adjustment module and the distribution module. Specifically, the working process of the evaluation module is as follows: when any one of the first-level node risk signal, the first-level reliability risk signal and the first-level expansion risk signal is received, it is determined that the current node load of the system is too high, an abnormal alarm signal is generated and sent to the adjustment module; when the second-level expansion risk signal is received, it is determined that the node load is high, an abnormal prompt signal is generated and sent to the adjustment module; when the node maintenance signal, the reliability maintenance signal and the expansion maintenance signal are received at the same time, it is determined that the node load is normal, a safety signal is generated and sent to the distribution module.

[0052] The adjustment module is used to receive the abnormal alarm signal and the abnormal prompt signal sent by the evaluation module, and according to the real-time state of the system, a load scheduling strategy is formulated and sent to the distribution module to ensure the normal operation of the system. Specifically, the scheduling strategy formulated by the adjustment module is as follows: when the abnormal alarm signal caused by the first-level node risk signal is received, the node load with the number i marked as existing risk is adjusted and migrated to other normal nodes; when the node load is high, the weight is reduced and the proportion of requests distributed to the node is reduced; when the abnormal alarm signal caused by the first-level reliability risk signal is received, the system data is regularly backed up; when the abnormal alarm signal caused by the first-level expansion risk signal and the abnormal prompt signal caused by the second-level expansion risk signal are received, the node cluster is horizontally expanded to meet the actual storage demand.

[0053] The distribution module is used to receive the security signal sent by the evaluation module and the load scheduling strategy sent by the adjustment module. When receiving the security signal, the current load data is analyzed and predicted, and the load distribution strategy is adjusted in advance, and the request is reasonably distributed according to the prediction result to avoid excessive load. Specifically, the work flow of the distribution module for adjusting the load distribution strategy is as follows:

[0054] S21, constructing a load redundancy value binary search tree: collecting load redundancy values, and the process of obtaining the load redundancy values is as follows. Specifically, the current load value of the node is brought into the formula wherein f is the current load value of the node, q is the redundancy requirement value of the node, is a correction factor, and is a preset proportion coefficient;

[0055] The first load redundancy value j is taken as a root node, and the second load redundancy value is sequentially inserted into the binary search tree. When a node is inserted, the size relationship between the node value and the current node is compared. If it is smaller than the current node, it is continuously inserted into the left subtree of the current node. If it is greater than the current node, it is continuously inserted into the right subtree of the current node. The process is repeated until all nodes are inserted into the binary search tree.

[0056] S22, in-order traversal of the load redundancy value binary search tree: in-order traversal is performed on the generated binary search tree, the left subtree is traversed first, then the root node is accessed, and finally the right subtree is traversed. The traversal result is an ascending order list of sorted load redundancy values.

[0057] S23, request distribution: according to the sorted load redundancy value sorting table, each node of the node cluster is assigned a weight value ki, ki is a number greater than 0, an ascending order list of node weight values is generated, the weight of the low weight node is increased, the proportion of request distribution to the node is improved, and the load of the node is improved. When a new request arrives at the dynamic load balancing module, the distribution module checks the current connection number and weight value of each node in the node cluster, and selects the node with the smallest connection number and weight value to process the request. The smallest connection number can ensure that the request is evenly distributed to each node, and the smallest weight value can proportionally distribute the request to different nodes.

[0058] In the embodiment of the present application, the evaluation module obtains the state of the current system through the received risk signal and maintenance signal, determines the load level of the system node, generates corresponding abnormal alarm signals, abnormal prompt signals and safety signals, and sends them to the adjustment module and the distribution module respectively; the adjustment module receives the abnormal alarm signals and abnormal prompt signals sent by the evaluation module, and formulates a reasonable load scheduling strategy according to the real-time state of the system, to ensure the availability and persistence of data; the distribution module analyzes and predicts the current load data through the safety signals sent by the evaluation module, adjusts the load distribution strategy in advance, reasonably distributes requests according to the prediction results, avoids excessive load of the system, and ensures the normal operation of the system.

[0059] The above merely describes the preferred embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can make equivalent replacements or changes to the technical solutions and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

[0060] The preferred embodiments of the present application disclosed above are only used to help explain the present application. The preferred embodiments do not describe all the details and limit the present application to the specific embodiments. Obviously, many modifications and changes can be made according to the content of the present application. The present application selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that the person skilled in the art can well understand and utilize the present application. The present application is limited by the claims and the entire scope and equivalents thereof.

Claims

1. A mass storage distributed storage system, characterized by: The system comprises a client module, a dynamic load balancing module, a node cluster and a monitoring module; the client module and the dynamic load balancing module are in bidirectional communication connection; the dynamic load balancing module and the node cluster are in unidirectional communication connection; the monitoring module and the dynamic load balancing module are in bidirectional communication connection; the monitoring module and the node cluster are in bidirectional communication connection; the node cluster is composed of multiple horizontally scalable nodes; the dynamic load balancing module comprises an evaluation module, an adjustment module and a distribution module; The client module is used for receiving the request of a user; The dynamic load balancing module is used for receiving the request sent by the client module, obtaining the risk signal and the maintenance signal of the current system from the monitoring module to determine the load condition of the system node, and dynamically distributing the request to multiple nodes in the node cluster; The node cluster is used for dispersively storing data on the nodes, using distributed storage technology to increase storage capacity and performance; The monitoring module is used for real-time monitoring and collecting the node performance data, reliability performance data and expansion performance data of the system, and performing in-depth analysis to generate the risk signal and the maintenance signal to determine the real-time state of the system; the working process of the monitoring module is as follows: S1, collecting the node performance data, reliability performance data and expansion performance data of the large-capacity distributed storage system; the node performance data includes CPU utilization, memory usage, disk I / O utilization and request response time; the reliability performance data includes mean time between failures and mean time to repair; The expansion performance data includes performance loss ratio and throughput growth ratio; S2, when any one of the following conditions occurs: the CPU utilization continuously exceeds 70% to 80%, the memory usage exceeds 80% to 90%, the disk I / O utilization exceeds 70% and the request response time continuously exceeds the set 500 ms threshold, a first-level node risk signal is generated; When the CPU utilization is less than or equal to 70%, the memory usage is less than or equal to 80%, the disk I / O utilization is less than or equal to 70% and the request response time is less than or equal to the set 500 ms threshold, a node maintenance signal is generated; when the average time to failure is less than a set threshold or the average repair time is greater than a set threshold a level I reliability risk signal is generated; when the average time to failure is greater than or equal to a set threshold and the average repair time is less than or equal to a set threshold a reliable maintenance signal is generated; when the performance loss ratio is higher than a set threshold and the throughput growth ratio is lower than a set threshold a level I expansion risk signal is generated; When the performance loss ratio is higher than a set threshold and the throughput growth ratio is higher than a set threshold a level II expansion risk signal is generated. When the performance loss ratio is lower than a set threshold And the throughput growth ratio is higher than a set threshold An extended maintenance signal is generated; all the above generated signals are sent to the dynamic load balancing module.

2. The large-capacity distributed storage system according to claim 1, wherein The evaluation module is used for analyzing and determining the received risk signal and maintenance signal, generating multiple signals and sending them to the adjustment module and the distribution module; The adjustment module is used for receiving the abnormal alarm signal and the abnormal prompt signal sent by the evaluation module, formulating a load scheduling strategy according to the real-time state of the system and sending it to the distribution module; The distribution module is used for receiving the safety signal sent by the evaluation module and the load scheduling strategy sent by the adjustment module, analyzing and predicting the current load data when the safety signal is received, adjusting the load distribution strategy in advance, and reasonably distributing the request according to the prediction result.

3. The mass distributed storage system of claim 2, wherein, The workflow of the evaluation module is as follows: when receiving any one of the level I node risk signal, the level I reliable risk signal and the level I expansion risk signal, it is determined that the current node load of the system is too high, an abnormal alarm signal is generated and sent to the adjustment module; when receiving the level II expansion risk signal, it is determined that the node load is high, an abnormal prompt signal is generated and sent to the adjustment module; when receiving the node maintenance signal, the reliable maintenance signal and the expansion maintenance signal at the same time, it is determined that the node load is normal, a safety signal is generated and sent to the distribution module.

4. The mass distributed storage system of claim 3, wherein, The scheduling strategy formulated by the adjustment module is as follows: when receiving the abnormal alarm signal caused by the level I node risk signal, the node load of the number i marked as existing risk is adjusted and migrated to other normal nodes; when the node load is high, the weight is reduced and the proportion of request distribution to the node is reduced; when receiving the abnormal alarm signal caused by the level I reliable risk signal, the system data is regularly backed up; when receiving the abnormal alarm signal caused by the level I expansion risk signal and the abnormal prompt signal caused by the level II expansion risk signal, the node cluster is horizontally expanded to meet the actual storage demand.

Images