A high-availability cloud storage method based on IP-SAN technology

Abstract

This invention discloses a high-availability cloud storage system based on IP-SAN technology. The high-availability cloud storage system includes: a computing service layer, a storage gateway layer, a data processing layer, and a storage array layer. Two sets of 10 Gigabit switches are respectively installed between the storage gateway layer and the data processing layer, and between the data processing layer and the storage array layer, with each set of 10 Gigabit switches serving as backups for the others. The data processing devices in the data processing layer are connected to the storage gateway layer and the storage array layer in an independent, serially connected manner. Each computing server in the computing service layer is interconnected with all storage gateways. Information systems deployed on the computing servers request storage file systems and storage space from the storage gateways as needed. The data processing devices in the data processing layer are placed between the storage gateways and the storage array in an independent, serially connected manner.

Description

Technical Field

[0001] This invention relates to the field of cloud storage systems, and more particularly to a high-availability cloud storage system based on IP-SAN technology. Background Technology

[0002] With the continuous development of domestically produced servers, computers, switches, disk arrays, operating systems, and other hardware and software products, the main functions and technical specifications of these products are also constantly improving. At the same time, the application of new technologies such as cloud computing, big data, and artificial intelligence in information systems has led to a rapid increase in the scale of various data resources, placing higher demands on the security, reliability, and efficiency of data storage systems.

[0003] In the construction of existing information systems, considering factors such as the cost, performance, and maturity of domestically produced equipment, IP-SAN is mainly used to meet the data storage and backup management needs of large-scale, multi-file systems. During implementation, there are mainly difficulties such as support for multiple file systems, insufficient high availability and high concurrency capabilities of data storage services, difficulty in dynamically expanding concurrency and storage space, and difficulty in quickly recovering from storage link failures.

[0004] With the continuous development of technologies such as domestic storage gateways, 10 Gigabit switches, and heterogeneous file systems, the storage architecture and access link scheduling methods of information systems are optimized, and the parallel integration of similar devices is expanded, enabling information systems to achieve high concurrency and high availability storage methods. Based on this, a high availability cloud storage method based on IP-SAN (IP-Storage Area Network) technology is proposed to meet the needs of the ever-evolving information system storage environment construction. Summary of the Invention

[0005] Purpose of the invention: The technical problem to be solved by the present invention is to provide a highly available cloud storage method based on IP-SAN technology, which addresses the shortcomings of the existing technology.

[0006] To address the aforementioned technical problems, this invention discloses a high-availability cloud storage system based on IP-SAN technology. The high-availability cloud storage system includes: a computing service layer, a storage gateway layer, a data processing layer, and a storage array layer.

[0007] Two sets of 10 Gigabit switches are set up between the storage gateway layer and the data processing layer, and between the data processing layer and the storage array layer. Each set of 10 Gigabit switches adopts horizontal virtualization technology to achieve network redundancy backup.

[0008] Each layer is connected sequentially via a 10 Gigabit IP network;

[0009] The data processing devices in the data processing layer are connected to the storage gateway layer and the storage array layer respectively in an independent, serially connected access manner;

[0010] Specifically, data is generated from the computing service layer and stored in the storage array; the entire process is called a storage link.

[0011] Furthermore, each layer is configured with at least two similar devices, and the devices in each layer are connected via a 10 Gigabit IP network;

[0012] Devices at the same level form a cluster to meet the needs of high concurrency and device redundancy.

[0013] Furthermore, each compute server in the compute service layer is interconnected with all storage gateways, thereby utilizing the storage resources and access services provided by the storage gateway cluster service. Information systems deployed on the compute servers request storage file systems and storage space from the storage gateways as needed.

[0014] The storage gateway layer consists of various devices forming a storage gateway cluster. It adopts distributed storage technology to achieve storage aggregation and load balancing. It is managed and scheduled uniformly through the storage gateway cluster service deployed on the storage gateway. It provides various file system storage services to the information system carried by the computing server, and provides services such as multi-replica verification storage, access path selection, backup and recovery management to the lower level.

[0015] Storage gateway cluster services include file system services and multi-replica access services;

[0016] The file system service includes pre-configured support for multiple commonly used distributed file systems such as NFS and S3. It can be enabled on demand and published to the computing service layer through configuration. It is mapped to a mountable local file system through the NFS / SMB standard file protocol, which simplifies the access of upper-layer applications to cloud data. At the same time, it provides local caching support, thereby realizing multi-file system support and optimized access performance.

[0017] The data multi-replica verification storage service is designed with a unified data storage service for different file system services. Each data storage operation generates multiple replicas based on the number of storage gateways, with each gateway selecting a storage path to achieve multi-replica storage on the disk array (allowing switching to an available path in case of switch, cable, or other failures). Simultaneously, the storage status feedback for each replica determines whether it is stored correctly. For abnormally stored replicas, the storage gateway cluster service automatically synchronizes the content of the normal data replicas, verifying and correcting the abnormal replicas, thereby improving storage reliability and data backup and recovery capabilities.

[0018] Furthermore, the data processing layer employs a "master-slave" cluster mode, with independent serial access placed between the storage gateway and the storage array. As a front-end processing gateway for the storage array, it provides multiple data storage links and proactive data processing services during the computing server's access to the storage array. This data processing layer is particularly suitable for housing data encryption devices, enabling unified encryption and decryption of stored data.

[0019] The number of devices can be dynamically expanded based on the amount of data accessed and the number of concurrent users during use.

[0020] The data processing layer: Based on the performance of data processing devices and the relationship of storage paths, some devices are grouped into a cluster to further improve the performance of local data access and processing, and meet the high concurrency needs of some information system applications;

[0021] Based on the reliability requirements of the data access link, some devices are used as redundant hot standby. When the data access link in use fails, the multi-path selection service can achieve failover, thereby improving the reliability of the link.

[0022] Furthermore, the storage array layer connects one or more disk array devices through a 10 Gigabit switch to form an independent redundant disk array (RAID). The RAID level is selected as needed, each disk array is divided into 1 to m logical volumes, and all logical volumes are allocated to n storage gateways. The storage gateway cluster service uniformly allocates and manages storage space and performs data backup and recovery according to the configuration.

[0023] Furthermore, in the cluster service of the storage gateway layer, the "link planning + detection and selection" approach is adopted to optimize the multipath selection service (multipath) link switching strategy, which includes four parts: link planning, primary link selection, backup link switching and primary link back-off, in order to improve the reliability of storage links and reduce fault recovery time.

[0024] The link planning involves mounting 1-m logical volumes (OSDs) under each storage gateway as needed, and allocating them according to file system type and storage space. It also pre-allocates the default storage path, preferred primary link, and backup link selection weights for each storage gateway, and distributes different primary links among the storage gateways.

[0025] The primary link selection is performed after the storage gateway cluster service is initially started. Each logical volume will detect available storage links as the primary link according to the link planning rules through the multi-path selection service. The selection results of each storage gateway are also considered to ensure that logical volumes under the same storage gateway do not use the same primary link.

[0026] The backup link switching is as follows: when the primary link of the logical volume is not faulty, the primary link will always be used as the storage communication link; when the primary link of the logical volume fails or is disconnected, the multipath selection service will re-detect the available storage links according to the backup link selection weight, and immediately switch the communication link of the logical volume to the backup link, and then upgrade the backup link to the primary link.

[0027] The primary link rollback: When the original faulty primary link recovers, the multipath selection service will not perform a rollback operation. Instead, it will add the link to the backup link of the corresponding logical volume and automatically set the backup link selection weight according to the rules. This weight will be used when switching backup links in the future, thereby avoiding frequent switching under abnormal conditions that could affect the stability of the entire storage cluster.

[0028] Furthermore, the necessary file system services are configured for each information system application in the computing service layer, thereby enabling all applications to use storage links and space uniformly and meeting the needs for efficient data access. For example, information system databases use the NFS file system by mounting, while cloud disk management applications use the S3 file system through service calls.

[0029] The data storage process corresponding to the aforementioned high-availability cloud storage system based on IP-SAN technology includes:

[0030] Step 1, Data Read / Write Request: The information system application on the computing server generates data objects and initiates data read / write requests to the storage gateway cluster;

[0031] Step 2, Data Replica Generation: After receiving the data object, the storage gateway cluster service creates a number of data object replicas equal to the number of storage gateways and distributes them to each storage gateway.

[0032] Step 3, Logical Volume Selection: After receiving the data object copy, each storage gateway selects the logical volume to be written to using an algorithm;

[0033] Step 4, Data Writing: After the logical volume receives the data object copy, it writes each data copy to the disk array according to the primary link selected by the logical volume, and synchronizes the storage status of each copy to ensure that the storage status of all copies is consistent. This completes one storage process from the creation of the data object to its writing.

[0034] Beneficial effects:

[0035] (1) The storage link architecture is planned in layers according to computing services, storage gateways, data processing and storage arrays, so that the number of devices can be flexibly expanded on demand at each layer, and the concurrent access and high availability capabilities can be improved;

[0036] (2) The storage links are connected through a 10 Gigabit IP network, which effectively reduces the technical difficulty of configuration and maintenance and reduces the cost of building the storage system;

[0037] (3) The storage gateway manages the disk array storage space and data access scheduling service in a unified manner, thereby meeting the different file system needs of various applications in the computing service layer, and can flexibly allocate the storage space and backup strategies of each file system through configuration;

[0038] (4) The storage gateway enables the selection of the primary storage link, the switching of the backup link and the return of the primary link. When a local device fails, the storage link can be switched quickly and the optimal link can be selected, thereby effectively improving the high concurrency and high availability of the entire cloud storage system. Attached Figure Description

[0039] Figure 1 This is a logical relationship diagram of the high-availability cloud storage method of the present invention.

[0040] Figure 2 This is a technical structure diagram of the high-availability cloud storage method of the present invention.

[0041] Figure 3 This is a flowchart of the data storage and multi-path selection process of the present invention.

[0042] Figure 4 shows the verification results of the high availability storage link switching of the present invention. Detailed Implementation

[0043] This invention is implemented during the construction of information systems based on domestically produced software and hardware environments. By optimizing the storage link structure, link selection strategy, and data access process, it can provide multi-file system support and reliable and efficient access services for various applications of upper-layer information systems, and can effectively reduce the overall cost of building the storage link.

[0044] The following describes the invention in further detail, using the construction of a typical information system storage link as an example and in conjunction with the accompanying drawings. Its application background is the construction of a storage link for a business information system to provide NFS, S3, and other file system services for upper-layer application software and to achieve high-availability storage for various types of data. The logical relationship of the related device connections is referenced below. Figure 1 Technical structure such as Figure 2 As shown, the data storage and multi-path selection process is as follows: Figure 3 As shown in Figure 4, the storage link switching verification results are shown in Figure 4.

[0045] Step 1: Plan the storage link architecture in layers according to computing services, storage gateway, data processing, and disk array. Based on the user access volume of the information system and the need for high availability and redundancy backup, each layer can be configured with 2 to n corresponding similar devices. The devices in each layer are connected through a 10 Gigabit IP network. Two sets of 10 Gigabit switches are set up between the three layers behind the storage link, and each set of 10 Gigabit switches serves as a backup for the other.

[0046] In this embodiment, the devices configured at each layer include: 5 computing servers, 4 10 Gigabit switches, 3 storage gateways, 6 disk arrays, and 5 data processing devices. The above devices refer to... Figure 1 Connect using logical relationships.

[0047] Step 2: The computing service layer connects to the storage gateway server cluster via the data storage network. Based on the information system service needs of each server, it requests the corresponding storage file system service and space from the storage gateway cluster service. For example, in this embodiment, the container cloud platform and database service request NFS file system service, while cloud disk management and file services can request S3 file system service, etc.

[0048] Step 3: The storage gateway layer forms a device cluster, which is managed and scheduled uniformly through the storage gateway cluster service. It provides various file systems to the computing service layer above and storage space management and access path selection services to the lower layer. The storage gateway cluster service includes functions such as file system management, storage space management, monitoring management, backup and recovery management, multi-path services, and rule management. Figure 2 As shown.

[0049] a) The storage gateway generates one or more file system services on demand and provides them to various information systems in the computing service layer, such as NFS, S3 and other file systems.

[0050] b) The storage gateway enables unified management of the storage space of all disk arrays, realizing the generation and distribution of data object copies, and the storage, access and backup management of various data resources.

[0051] c) The storage gateway provides multi-path selection services, selects the optimal storage link according to the link selection strategy, or switches to another available path after the switch, cable, etc. fail.

[0052] Step 4: The processing devices in the data processing layer adopt a "master-backup" cluster mode, independently connected in series, and placed between the storage gateway layer and the storage array layer. According to the storage path relationship, they provide data processing services for the data passing through. Some devices can be clustered to improve local access performance; others can be used as redundancy to achieve failover in case of failure, improving the reliability of access services. For example, in this embodiment, the data processing layer uses 5 data encryption processing devices to perform unified encryption and decryption processing on the stored data.

[0053] Step 5: The storage array layer connects one or more disk array devices through a 10 Gigabit switch to form an independent redundant disk array (RAID). The RAID level is selected as needed. Each disk array can be divided into 1 to n logical volumes (OSDs). All logical volumes are allocated to n storage gateways. The storage gateway cluster service is responsible for the unified allocation and management of storage space and performs data backup and recovery according to the configuration.

[0054] In this embodiment, a cluster storage space is composed of 6 disk arrays connected by 2 10 Gigabit switches. The total storage space capacity is 400T, using RAID5 level. Each disk array is divided into 6 logical volumes, for a total of 36 logical volumes (OSD0 to OSD35). The 12 logical volumes per disk array are evenly distributed to 3 storage gateways.

[0055] Step 6: Configure storage link switching policies in the storage gateway cluster service, including link planning, primary link selection, backup link switching, and primary link fallback, to improve the reliability of storage links.

[0056] a) Link planning. Mount 1-m logical volumes (OSDs) under each storage gateway as needed, and allocate them according to file system type and storage space. Pre-allocate the default storage path, preferred primary link, and backup link selection weights for each storage gateway, and distribute the use of different primary links among the storage gateways;

[0057] b) Primary Link Selection. After the storage gateway cluster service is initially started, each logical volume will detect available storage links as the primary link according to the link planning rules through the multi-path selection service, and combine the link selection results corresponding to each storage gateway to ensure that logical volumes under the same storage gateway do not use the same primary link.

[0058] c) Backup link switching. When the primary link of a logical volume is not faulty, it will always be used as the storage communication link. When the primary link of a logical volume fails or is disconnected, the multipath selection service will re-detect available storage links according to the backup link selection weight, immediately switch the communication link of the logical volume to the backup link, and then upgrade the backup link to the primary link.

[0059] d) Primary link fallback. When the original faulty primary link recovers, the multipath selection service will not perform a fallback operation. Instead, it will add the link to the backup link of the corresponding logical volume and automatically set the backup link selection weight according to the rules. This will be used during subsequent backup link switching, thereby avoiding frequent switching under abnormal conditions that could affect the stability of the entire storage cluster.

[0060] In this embodiment, for example, OSD2, OSD13 and OSD25 all selected device 2 as the primary link. When no failure occurs, data objects are stored through device 2. When device 2 fails, the multipath selection service switches device 3 as the backup link, so that the storage service can continue to be available. When device 2 recovers, the multipath selection service will include device 2 in the backup storage link.

[0061] Step 7: After completing the above deployment and configuration, the data storage process mainly implements data read and write operations through four steps: data read / write requests, data copy generation, logical volume selection, and data read / write. In this embodiment, a file upload using a cloud disk management application is taken as an example to specifically illustrate the data storage process, such as... Figure 3 As shown.

[0062] ① Data read / write requests. Information system applications on the compute server generate data objects and initiate data read / write requests to the storage gateway cluster. For example, a cloud disk management application running on compute server 1 initiates a file upload storage request, generating a data object to be processed.

[0063] ② Data Replica Generation. After receiving a data object, the storage gateway cluster service creates a number of data object replicas equal to the number of storage gateways and distributes them to each storage gateway. In this embodiment, the data object generated by the computing server will create three data replicas, which will be distributed to storage gateways 1 through 3 respectively.

[0064] ③ Logical Volume Selection. After receiving a copy of the data object, each storage gateway selects the logical volume to be written to using an algorithm. For example, in this data object storage process, storage gateway 1 selects OSD2, storage gateway 2 selects OSD13, and storage gateway 3 selects OSD25.

[0065] ④ Data Writing. After the logical volume receives a copy of the data object, it writes each data copy to the disk array according to the primary link selected by the logical volume. For example, in this data object storage, OSD2, OSD13, and OSD25 all selected device number 2 as the primary link. After the data object is processed, it is written to the corresponding logical volume, thus completing a storage process from data object creation to data object writing.

[0066] The following section further verifies the information system storage link constructed in this embodiment by combining high-concurrency data writing and storage link device abnormality scenarios.

[0067] a) Check the link allocation in the initial state. View the initial link allocation through the storage gateway cluster service, such as... Figure 4a As shown: Initial state query results: Storage gateway 1 uses encryption device 1 to mount 14 OSDs; Storage gateway 2 uses encryption device 1 to mount 5 OSDs, encryption device 2 to mount 1 OSD, encryption device 3 to mount 6 OSDs, and encryption device 5 to mount 2 OSDs; Storage gateway 3 uses encryption device 1 to mount 8 OSDs and encryption device 5 to mount 6 OSDs.

[0068] Table 1. Link Allocation in Initial State

[0069] equipment Storage Gateway No. 1 Storage Gateway No. 2 Storage Gateway No. 3 Encryption device No. 1 14 5 8 Encryption device No. 2 1 Encryption device No. 3 6 Encryption device No. 4 Encryption device No. 5 2 6

[0070] b) Verify link adjustment under high concurrency. Simulate 300 concurrent write operations on the compute server, writing files of random sizes from 1MB to 100MB each time, continuously writing approximately 200GB of data. Then, observe the link adjustment under high concurrency through the storage gateway cluster service. Figure 4b The high-concurrency status query results show that the storage gateway cluster service adjusts the load balancing of storage links based on high concurrency, data copy writing and synchronization, and changes the encryption machine and OSD mounting allocation used by the storage gateway.

[0071] Table 2. Verification of High-Concurrency Downlink Adjustment

[0072] equipment Storage Gateway No. 1 Storage Gateway No. 2 Storage Gateway No. 3 Encryption device No. 1 6 8 Encryption device No. 2 8 Encryption device No. 3 6 Encryption device No. 4 8 Encryption device No. 5 6

[0073] c) Verify link adjustment during device failure. Simulate a device failure by disconnecting the network connection of encryption device #2. Then, check the link adjustment under high concurrency using the storage gateway cluster service again. Figure 4c The device fault status query results show that the storage links of the storage gateway using encryption device No. 2 have all been switched to other encryption devices, achieving high availability operation of the entire storage system.

[0074] Table 3. Link Adjustment During Verification Equipment Failure

[0075] equipment Storage Gateway No. 1 Storage Gateway No. 2 Storage Gateway No. 3 Encryption device No. 1 6 8 Encryption device No. 2 Encryption device No. 3 4 6 Encryption device No. 4 1 8 Encryption device No. 5 3 6

[0076] In summary, the storage link of a business information system constructed using the solution of this invention can effectively address issues such as insufficient support for multiple file systems, inadequate high availability and high concurrency capabilities of data storage services, difficulty in dynamically expanding concurrency and storage space, and difficulty in quickly recovering from storage link failures.

[0077] This invention provides a highly available cloud storage system based on IP-SAN technology. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A high-availability cloud storage system based on IP-SAN technology, characterized in that, The high-availability cloud storage system includes: a computing service layer, a storage gateway layer, a data processing layer, and a storage array layer; Two sets of 10 Gigabit switches are set up between the storage gateway layer and the data processing layer, and between the data processing layer and the storage array layer, with each set of 10 Gigabit switches serving as backups for the others.

2. The high-availability cloud storage system based on IP-SAN technology according to claim 1, characterized in that, The data processing devices in the data processing layer are connected to the storage gateway layer and the storage array layer respectively in an independent, serially connected access manner; Each computing server in the computing service layer is interconnected with all storage gateways. Information systems deployed on the computing servers apply to the storage gateways for storage file systems and storage space as needed. The data processing devices in the data processing layer are placed between the storage gateway and the storage array in an independent, serially connected manner.

3. A high-availability cloud storage system based on IP-SAN technology according to claim 1, characterized in that, Devices at each layer are connected via a 10 Gigabit IP network, and devices at the same layer form a cluster.

4. A high-availability cloud storage system based on IP-SAN technology according to claim 1, characterized in that, The storage gateway layer consists of various devices forming a storage gateway cluster. The storage gateway cluster service deployed on the storage gateway is used for unified management and scheduling. It provides multiple file systems for the information system carried by the computing server above and implements multi-replica verification storage below.

5. The high-availability cloud storage system based on IP-SAN technology according to claim 4, characterized in that, Storage gateway cluster services include file system services and multi-replica access services; The file system service includes support for more than one commonly used distributed file system, which can be enabled on demand and published to the computing service layer through configuration. It is mapped to a mountable local file system through the NFS / SMB standard file protocol.

6. A high-availability cloud storage system based on IP-SAN technology according to claim 5, characterized in that, The data multi-replica verification storage service is designed as a unified data storage service for different file system services. Each data storage generates multiple replicas according to the number of storage gateways. Each storage gateway selects a storage path to achieve multi-replica storage on the disk array. At the same time, based on the storage feedback status of each replica, it is determined whether each replica is stored correctly. For abnormal replicas, the storage gateway cluster service automatically synchronizes the content of normal data replicas to realize the verification and correction of abnormal replicas.

7. A high-availability cloud storage system based on IP-SAN technology according to claim 6, characterized in that, In the cluster service of the storage gateway layer, a combination of link planning and detection selection is adopted to optimize the link switching strategy of the multi-path selection service, which includes four parts: link planning, primary link selection, backup link switching and primary link back-off.

8. The high-availability cloud storage system based on IP-SAN technology according to claim 2, characterized in that, The data processing layer groups some devices into a cluster based on the performance of the data processing equipment and the relationship of storage paths; and uses some devices as redundant hot standby. When a failure occurs in the data access link in use, the multi-path selection service can achieve failover.

9. The high-availability cloud storage system based on IP-SAN technology according to claim 8, characterized in that, The storage array layer connects one or more disk array devices through a 10 Gigabit switch to form an independent disk redundancy array. The independent disk redundancy array level can be selected as needed. Each disk array is divided into 1 to m logical volumes, and all logical volumes are allocated to n storage gateways. The storage gateway cluster service uniformly allocates and manages storage space and performs data backup and recovery according to the configuration.

10. The high-availability cloud storage system based on IP-SAN technology according to claim 5, characterized in that, The distributed file system includes: a network file system and a simple storage service.

Images