Storage server

Abstract

The utility model discloses a storage server, relates to the computer field, said storage server includes at least one computing module, central backplane, sharing module, wherein each computing module includes mainboard, mainboard includes central processing unit and memory, sharing module includes the storage equipment that is used for providing the service for at least one computing module, the central processing unit includes the central processing unit and the memory. And each calculation module is connected with the sharing module through the middle backboard.

Description

Technical Field

[0001] This utility model embodiment is applied in the field of computers, and more specifically, relates to a storage server. Background Technology

[0002] In traditional server architectures, components (such as motherboards, backplanes, I / O expansion boards, etc.) are tightly coupled, making the process of adding or replacing components complex and difficult.

[0003] Therefore, existing technologies suffer from the problem of low flexibility in expansion due to the strong interdependence of hardware components. Utility Model Content

[0004] This utility model provides a storage server to at least solve the technical problem of low scalability.

[0005] According to one embodiment of the present invention, a storage server is provided, comprising: at least one computing module, a central backplane, and a shared module, wherein each computing module includes a motherboard, the motherboard includes a central processing unit and memory, the shared module includes a storage device for providing services to at least one computing module, and each computing module is connected to the shared module through the central backplane.

[0006] Optionally, as an optional embodiment, the storage server further includes at least one dedicated module, each dedicated module being used to provide services to its corresponding computing module, and each computing module being connected to its corresponding dedicated module.

[0007] Alternatively, as an optional embodiment, each computing module is connected to its corresponding dedicated module via a high-speed connector.

[0008] Optionally, as an optional embodiment, each dedicated module includes at least one of the following: a dedicated management module for managing the computing modules corresponding to each dedicated module; an input / output card for providing input / output services to the computing modules corresponding to each dedicated module; and a high-speed computer expansion device for providing expansion services to the computing modules corresponding to each dedicated module.

[0009] Optionally, as an optional embodiment, when the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the central backplane through its corresponding first integrated circuit interconnect bus, and the central backplane is connected to the storage device through a second integrated circuit interconnect bus. The first connection channel established between the central processing unit in each computing module and the storage device through the central backplane is the management channel used by each computing module to manage the storage device.

[0010] Optionally, as an alternative embodiment, when the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the storage device via a high-speed connector, and the second connection channel between the central processing unit in each computing module and the storage device is a data transmission channel used by the storage device to provide data storage services for each computing module.

[0011] Optionally, as an optional embodiment, if the number of computing modules in the storage server is greater than the second threshold, the shared module further includes: a management module for managing at least one computing module; and a switching module, wherein the switching module includes at least one switching expansion card, each switching expansion card is configured with at least one switch chip, and the second threshold is greater than or equal to the first threshold.

[0012] Optionally, as an alternative embodiment, if the number of computing modules in the storage server is greater than the second threshold, the central processing unit in each computing module is connected to the switching module through a central backplane, and the switching module is connected to each storage device in the storage device.

[0013] Optionally, as an alternative embodiment, the central processing unit in each computing module is connected to at least one switching expansion card via a central backplane, and the at least one switching expansion card is connected to each storage device in the storage device via a high-speed serial computer expansion bus.

[0014] Optionally, as an alternative embodiment, the shared module further includes a shared input / output card for providing input / output services to at least one computing module.

[0015] This invention solves the problem of large-scale hardware redesign caused by platform upgrades in the prior art by dividing the storage server into at least one computing module, a central backplane, and a shared module. It also overcomes the shortcomings of traditional architectures in terms of flexibility, scalability, and management efficiency, and achieves the technical effect of improving the scalability and flexibility of the storage server. Attached Figure Description

[0016] To more clearly illustrate the embodiments of this utility model, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this utility model. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A structural diagram of a storage server provided in an embodiment of this utility model;

[0018] Figure 2A connection diagram of a computing module and a sharing module provided for an embodiment of this utility model;

[0019] Figure 3 Another connection diagram of the computing module and the sharing module provided in this embodiment of the utility model;

[0020] Figure 4 An exploded view of a storage service server provided in an embodiment of this utility model;

[0021] Figure 5 An exploded view of another storage service server provided in an embodiment of this utility model. Detailed Implementation

[0022] The technical solutions of the present utility model will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present utility model, and not all embodiments. Based on the embodiments of the present utility model, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present utility model.

[0023] It should be noted that in the description of this utility model, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this utility model are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0024] An embodiment of this utility model provides a storage server, and the contents related to this utility model are explained herein:

[0025] 1) Computing module, central backplane, and shared module:

[0026] Computing Module: As the core computing unit, each module includes a motherboard, a central processing unit (CPU), and memory. Different CPU models and memory capacities can be selected according to different performance requirements to adapt to diverse computing needs.

[0027] Central backplane: It acts as a bridge, connecting the computing module and the sharing module, managing the transmission of data flow and control signals, and ensuring high-speed and reliable communication.

[0028] Shared modules: These include storage devices that can be accessed by multiple computing modules, enabling data sharing and optimized resource utilization, thereby enhancing the overall storage capacity and availability of the system.

[0029] 2) Dedicated Module:

[0030] Each computing module is also connected to at least one dedicated module, which provides it with specific services such as management, enhanced input / output capabilities, or high-speed scaling services, ensuring that each node receives optimal resource allocation based on its specific task.

[0031] 3) High-speed connector:

[0032] Using high-performance connectors (such as high-speed PCIe connectors) to connect computing modules and dedicated modules ensures data transmission speed and stability, making it suitable for high-performance computing environments.

[0033] 4) Diverse functions of the exclusive module:

[0034] Dedicated modules may include management modules, input / output cards, or high-speed computer expansion devices. These devices correspond one-to-one with computing modules, providing customized services and support to meet individual computing task needs.

[0035] 5) Communication architecture for servers with a low number of nodes:

[0036] When the number of computing modules does not reach the first threshold, the CPU and the central backplane are connected through the first integrated circuit interconnect bus, such as the Peripheral Component Interconnect Express (PCIe) bus. The backplane is then connected to the storage device through the second integrated circuit interconnect bus (such as PCIe Gen5), forming an efficient management channel and simplifying the communication path.

[0037] 6) Optimization of data transmission channels:

[0038] With a low number of nodes, the CPU can directly establish a data transmission channel with the storage device through a high-speed connector, ensuring high speed and low latency for data read and write operations and improving the performance of the storage server.

[0039] 7) Modular expansion of high-node-count servers:

[0040] When the number of computing modules exceeds the second threshold, a management module and a switching module need to be introduced. The former is responsible for overall monitoring and control, while the latter (including switching expansion cards and switch chips) supports large-scale data exchange, ensuring communication efficiency and stability in a multi-node environment.

[0041] 8) Interconnection solutions for large-scale systems:

[0042] In high-node-count servers, each CPU interacts indirectly with the switching module through a central backplane. The switching module connects all storage devices to form a distributed storage network, which enhances data exchange capabilities and the overall scalability of the system.

[0043] 9) High-speed serial computer expansion bus:

[0044] It adopts high-speed serial bus technology to realize direct connection between computing modules, switching expansion cards and storage devices, which further improves data transmission speed and communication capacity, making it suitable for ultra-large-scale storage server deployment.

[0045] 10) Sharing of input / output functions:

[0046] In addition to storage devices, shared modules also include shared input / output cards, allowing multiple computing modules to share input / output (I / O) resources, improving the processing efficiency of I / O-intensive tasks, reducing overall costs, and making them suitable for large-scale data processing centers.

[0047] As an optional embodiment, the above-mentioned storage server, such as Figure 1 As shown, it includes: at least one computing module ( Figure 1 Taking the example of two computing modules (102 and 104), a central backplane 106, and a shared module 108, each computing module includes a motherboard, which includes a central processing unit and memory. The shared module includes a storage device 110 for providing services to at least one computing module. Each computing module is connected to the shared module through the central backplane.

[0048] Optionally, the aforementioned computing modules may, but are not limited to, represent one or more processing units of a storage server. Each computing module includes a motherboard, which is the core computing component in the storage server, responsible for data processing and execution of computing tasks. The number of computing modules can be adjusted according to the server's design scale and performance requirements.

[0049] It should be noted that the aforementioned central backplane may include, but is not limited to, a central communication platform for connecting computing modules and shared modules. It manages and coordinates data flow and signal transmission between the various modules, ensuring high-speed and reliable information exchange between them.

[0050] Furthermore, the aforementioned shared module can be, but is not limited to, the data-sharing portion of the storage server, including storage devices and other possible shared resources such as network interfaces or power supplies. Its existence allows multiple computing modules to access the same stored data, improving resource utilization efficiency and data access convenience.

[0051] Optionally, the aforementioned motherboard may, but is not limited to, be used to house the main hardware components of the computing module, such as the central processing unit (CPU), memory, and input / output (I / O) interfaces. It is the core of the computing module, responsible for connecting and managing all internal components, as well as communicating with other modules.

[0052] Optionally, the aforementioned central processing unit may, but is not limited to, represent the core computing component in the storage server, responsible for executing instruction sets and processing data, and is a key manifestation of the server's computing power.

[0053] Furthermore, the aforementioned memory may be, but is not limited to, volatile memory used to temporarily store currently processed data and instructions, including random access memory (RAM). The size of the memory directly affects the server's processing speed and multitasking capabilities.

[0054] It should be noted that the aforementioned storage devices may include, but are not limited to, hard disk drives, solid-state drives, and non-volatile memory devices, which provide non-volatile data storage space, enabling computing modules to persistently store and access data.

[0055] By adopting the embodiments of this utility model, the storage server is divided into at least one computing module, a central backplane and a shared module, which solves the problem of large-scale hardware redesign caused by platform upgrades in the prior art. At the same time, it overcomes the shortcomings of traditional architecture in terms of flexibility, scalability and management efficiency, and achieves the technical effect of improving the scalability and flexibility of the storage server.

[0056] As an optional embodiment, the storage server also includes at least one dedicated module, each dedicated module providing services to its corresponding computing module, and each computing module being connected to its corresponding dedicated module.

[0057] Optionally, the aforementioned dedicated module may, but is not limited to, refer to a set of hardware components specifically designed to provide particular services to a computing module. These services may include, but are not limited to, managing and monitoring the operating status of the computing module, providing additional input / output capabilities (such as connecting additional hard drives, network interface cards, etc.), or providing high-speed expansion services to the computing module to enhance its ability to process specific types of data.

[0058] It should be noted that "each computing module is connected to its corresponding dedicated module" means that there is a one-to-one connection between the computing module and the dedicated module. This means that each computing module has its own dedicated module. This setup ensures efficient resource allocation and management, and also facilitates customized service provision based on the characteristics and task requirements of the computing module.

[0059] Optionally, the above services may include, but are not limited to, the following types: Management services: Management modules contained in the dedicated module are used to detect and control the status and performance of the corresponding computing module, providing maintenance functions such as initialization, fault detection, and software updates. Input / output (I / O) services: Provided through the I / O card in the dedicated module, used to enhance the data input and output capabilities of the computing module, such as supporting more external device connections or providing higher bandwidth I / O interfaces to meet the needs of data-intensive applications. High-speed expansion services: The dedicated module may also contain high-speed expansion devices, such as additional GPUs or other accelerators, to improve the computing module's computing capabilities in specific fields, such as graphics processing and artificial intelligence inference.

[0060] In this embodiment of the utility model, the dedicated module provides personalized services to each computing module, including management functions, input / output services, or high-speed expansion services. In other words, each computing node can obtain the most suitable resources according to its specific needs and task nature, thereby improving resource utilization efficiency and task execution speed.

[0061] As an optional embodiment, each computing module is connected to its corresponding dedicated module via a high-speed connector.

[0062] It should be noted that a high-speed connector is a high-performance physical interface used to provide high-speed data transmission and communication between computing modules and dedicated modules. It can include, but is not limited to, high-speed PCIe (Peripheral Component Interconnect Express) connectors or other similar high-speed data link technologies to ensure fast and stable data transmission and meet the requirements of high-performance computing environments.

[0063] By using a high-speed connector to connect the computing module and the dedicated module, this utility model achieves high-speed data transmission and processing, especially in tasks with high bandwidth requirements, effectively improving the overall responsiveness and processing speed of the system.

[0064] As an optional embodiment, each dedicated module includes at least one of the following:

[0065] A dedicated management module for managing the computing modules corresponding to each dedicated module;

[0066] Input / output cards used to provide input / output services for the computing modules corresponding to each dedicated module;

[0067] High-speed computer expansion devices used to provide extended services for each dedicated module's corresponding computing module.

[0068] Optionally, the aforementioned dedicated management module may, but is not limited to, represent: local management functions of the computing module: each dedicated management module is responsible for detailed management of its corresponding computing module, including but not limited to monitoring CPU and memory status, fault detection and recovery, configuration settings, etc., to ensure the stable operation and high performance of the computing module.

[0069] It should be noted that the aforementioned input / output cards may include, but are not limited to: I / O accelerators: I / O interfaces or protocols that support high-speed data transmission, such as Serial Advanced Technology Attachment (SAS), Serial Attached Small Computer System Interface (SATA), and Non-Volatile Memory Express (NVMe), to improve data read and write speeds. Network adapters: Equipped with network interfaces, such as Ethernet and InfiniBand, to support high-speed communication between the computing module and other hardware components or external networks. General-purpose I / O expansion: Providing additional I / O ports or controllers, such as Universal Serial Bus (USB) and Peripheral Component Interconnect Express (PCIe), for connecting peripherals or other I / O devices.

[0070] Furthermore, the aforementioned high-speed computer expansion devices may include, but are not limited to: high-speed storage expansion: such as Non-Volatile Memory Express Solid State Drive (NVMe SSD), PCIe SSD, etc., significantly improving storage read and write performance. Graphics processing unit: i.e., GPU, used to accelerate computationally intensive applications, such as AI training, video rendering, etc. Field-programmable gate array: i.e., FPGA, which can be used for customized acceleration of logic operations or execution of specific algorithms. Network expansion devices: such as 100Gb / s Ethernet cards, used to enhance the network communication capabilities of computing modules.

[0071] By adopting the embodiments of this utility model, the dedicated module provides a variety of service options, including management, input / output and high-speed expansion, which can be selected according to the specific needs of the computing node, realizing the customized service capabilities of the storage server and meeting the special needs of different application scenarios.

[0072] As an optional embodiment, when the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the central backplane through its corresponding first integrated circuit interconnect bus. The central backplane is connected to the storage device through a second integrated circuit interconnect bus. The first connection channel established between the central processing unit in each computing module and the storage device through the central backplane is the management channel used by each computing module to manage the storage device.

[0073] It should be noted that the first threshold mentioned above refers to the preset upper limit of the number of computing modules. When the actual number of computing modules is lower than this threshold, the system adopts different communication architectures and resource management strategies.

[0074] Optionally, the aforementioned first integrated circuit interconnect bus may be, but is not limited to, a PCI Express bus (abbreviated as PCIe), a high-speed point-to-point serial connection standard used to connect the CPU and the backplane, providing high-speed data transmission and control signal communication.

[0075] It should be noted that the aforementioned center backplane is one of the core components of the storage server, serving as a bridge between the computing module and the storage device. Through it, different modules can perform efficient data exchange and signal transmission.

[0076] Optionally, the aforementioned second integrated circuit interconnect bus also uses the PCI Express bus, but may represent a higher version or better transmission characteristics, for data communication between the backplane and the storage device, ensuring high speed and low latency data transmission.

[0077] Furthermore, the aforementioned first connection channel is based on the first integrated circuit interconnect bus, that is, a high-speed channel established between the CPU and the backplane via the PCIe bus, used for the transmission of management and control signals.

[0078] Optionally, the aforementioned management channel is dedicated to the CPU's monitoring, configuration, and management of storage devices. Unlike the data transmission channel, the management channel primarily transmits control information and instructions to ensure the normal operation and efficient management of the storage devices.

[0079] In this embodiment of the invention, when the number of computing modules is small, the CPU is directly connected to the central backplane via the first integrated circuit interconnect bus (such as PCIe), and the central backplane is then connected to the storage device via the second integrated circuit interconnect bus, forming an efficient management channel, which simplifies the system architecture and improves the management efficiency of the storage device.

[0080] As an optional embodiment, when the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the storage device via a high-speed connector, and the second connection channel between the central processing unit in each computing module and the storage device is a data transmission channel used by the storage device to provide data storage services for each computing module.

[0081] Optionally, the statement that the number of computing modules is less than the first threshold can be used, but is not limited to, to represent a state in storage server design where the system is small and does not require a large number of computing nodes to maintain or improve its computing power or data processing efficiency. In this case, the number of computing modules is relatively limited, possibly to meet the computing needs of small businesses and laboratories, or to streamline the use of computing resources in specific application scenarios.

[0082] It should be noted that the aforementioned "central processing unit (CPU) connecting to storage devices via high-speed connectors" may include, but is not limited to, the process by which the CPU establishes a high-speed connection directly with the storage device when the number of computing modules is below a first threshold, in order to improve data read / write speed and reduce communication latency. High-speed connectors typically refer to physical interfaces that provide high-performance data transfer, such as PCI Express (PCIe) connectors, which enable direct high-speed communication between the CPU and storage devices.

[0083] Furthermore, the aforementioned second connection channel can refer to, but is not limited to, a dedicated data transmission channel established between the CPU and storage devices in server architectures with a relatively small number of computing modules. Unlike channels primarily responsible for control and management, the second connection channel focuses on high-speed data read and write, serving as a direct link at the data transmission layer and ensuring efficient and immediate data processing.

[0084] Optionally, the data transmission channel of the aforementioned data storage service may, but is not limited to, represent the high-speed communication path relied upon by the storage device when providing data storage services to various computing modules in a low-node-count server environment. The design goal of this channel is to maximize data transmission rate, minimize latency, and ensure that the server performs well when performing data-intensive tasks, such as database queries, data mining, and big data analytics.

[0085] In other words, in this embodiment, for storage servers with fewer controllers (e.g., dual controllers), the storage server can be split from front end to back end, but is not limited to. A shared module is defined at the front end, and the shared device is a hard drive or an I / O card. Sharing means that any node can access it and any node can perform read and write operations on the hard drive. The shared module's devices are connected to the computing module through a central backplane. The front end of the central backplane connects to the shared devices, and the back end connects to the computing module. All high-speed traces or low-speed management buses of the computing module converge on the central backplane and are then distributed to the various devices.

[0086] Optionally, behind the shared module is the computing module, i.e., the motherboard, which mainly includes the CPU, memory, and other key computing-related chips, responsible for core computing functions. Separating the computing module allows for compatibility with different CPU platforms. One end of the computing module connects to the backplane via a high-speed connector, and the other end connects to a dedicated I / O module. A dedicated module means it belongs exclusively to this motherboard. The dedicated device can be the controller's rear I / O card or any PCIe device, and it is unique to a single node. Dedicated devices also include the management module. For storage servers with few controllers, the management module is a dedicated device but belongs to this motherboard. It can be an external network port, such as a USB port or a serial port, providing engineers with an external interface to configure the device. In this way, four modules form an integrated disk controller server. Different types of servers can reuse the shared and dedicated modules, and the computing module can select different CPU platforms according to the performance requirements of different products.

[0087] It's important to note that for dedicated I / O modules, the compute module is a reusable part across different machine models. Shared modules require modification based on the number of nodes. The function of shared modules is to allow connected devices to be shared across all nodes in the system. For example, in a dual-controller node, the shared devices are typically hard drives, which can be, but are not limited to, dual-port Serial Attached SCSI Solid State Drives (SAS SSDs), Serial Attached SCSI Hard Disk Drives (SAS HDDs), or NVMe SSDs. One port corresponds to one device, thus achieving device sharing. For instance, in storage servers with fewer controllers (e.g., dual controllers), such as... Figure 2As shown, computing module 102 is connected to storage device 110 (such as hard disk) in shared module 108 via high-speed serial computer expansion bus 101, and computing module 104 is connected to storage device 110 in shared module 108 via high-speed serial computer expansion bus 103. Computing module 102 is also connected to central backplane 106 via integrated circuit interconnect bus 105, and computing module 104 is also connected to central backplane 106 via integrated circuit interconnect bus 107. Central backplane 106 is connected to storage device 110 via integrated circuit interconnect bus 109. In this way, the two computing modules are shared from the business link and management link.

[0088] By employing this embodiment of the invention, when the number of computing modules is small, the CPU establishes a data transmission channel directly with the storage device through a high-speed connector, which effectively improves the data reading and writing speed and reduces data transmission latency. It is particularly suitable for data-intensive task processing, such as database queries and big data analysis.

[0089] As an optional embodiment, if the number of computing modules in the storage server is greater than the second threshold, the shared module further includes: a management module for managing at least one computing module; and a switching module, wherein the switching module includes at least one switching expansion card, each switching expansion card is configured with at least one switch chip, and the second threshold is greater than or equal to the first threshold.

[0090] It should be noted that the second threshold mentioned above can be, but is not limited to, defining the limit on the number of computing modules when the storage server enters a large-scale deployment mode. When the number of computing modules exceeds this threshold, the system needs to introduce more complex management and data exchange mechanisms to optimize communication efficiency and resource coordination.

[0091] Optionally, the aforementioned management module may, but is not limited to, represent a dedicated hardware component responsible for detecting and controlling the overall operating status of the storage server. It can centrally manage all computing modules, enabling automated task scheduling, fault diagnosis and repair, and optimized resource allocation, ensuring the server's efficiency and reliability in large-scale deployments.

[0092] Furthermore, the aforementioned switching module can be, but is not limited to, a component in a storage server used to support high-speed data transmission and communication between multiple nodes. It is designed to build a robust network matrix by configuring at least one switching expansion card, each card carrying at least one switch chip, enabling computing modules to freely and efficiently exchange data, unaffected by an increase in the number of nodes.

[0093] It should be noted that the aforementioned switching expansion cards may include, but are not limited to, a series of highly customized hardware cards. These cards are inserted into the switching module and, with the help of the built-in switch chip, establish data communication paths in a multi-node environment. These cards can dynamically adjust connections according to system requirements, ensuring the flexibility and speed of data transmission.

[0094] Furthermore, the aforementioned switch chip can be, but is not limited to, a core component of the switching expansion card, responsible for packet forwarding and communication protocol processing. By supporting advanced switching functions (such as flow control, multipath routing, and virtualization), it enables efficient data exchange in complex network architectures, ensuring that storage servers maintain high performance even with a large number of compute nodes.

[0095] By adopting the embodiments of this utility model, when the number of computing modules exceeds the second threshold, a management module and an exchange module are introduced, which can realize the effective management and data exchange of a large number of computing nodes, especially in cluster and data center environments, improving management efficiency and data transmission capabilities.

[0096] As an optional embodiment, when the number of computing modules in the storage server is greater than the second threshold, the central processing unit in each computing module is connected to the switching module through a central backplane, and the switching module is connected to each storage device in the storage device.

[0097] Optionally, the above-mentioned "connected to the switching module via a central backplane" may not be limited to describing the network structure of how the computing module accesses the storage server. When there are a large number of nodes, the central backplane acts as an intermediate layer, responsible for routing communication requests from the central processing unit of each computing module to the switching module. The switching module then connects to various storage devices, forming a distributed storage network to support large-scale data exchange and storage services.

[0098] In this embodiment of the invention, when there are a large number of computing modules, the computing nodes are connected to the switching modules through a central backplane, and then establish data communication with all storage devices through the switching modules. This achieves efficient exchange and storage of large-scale data, and enhances the communication performance and data processing capabilities of the system in a multi-node environment.

[0099] As an optional embodiment, the central processing unit in each computing module is connected to at least one switching expansion card via a central backplane, and the at least one switching expansion card is connected to each storage device in the storage device via a high-speed serial computer expansion bus.

[0100] Optionally, the aforementioned high-speed serial computer expansion bus may, but is not limited to, represent a high-speed peripheral component interconnect express (PCIe) bus. It is a standard interface used for high-speed data transmission between internal hardware components, achieving high-bandwidth and low-latency communication through serial communication technology. In this embodiment, the high-speed serial computer expansion bus acts as a channel for high-speed data exchange between the computing module CPU and the switching expansion card and storage devices, ensuring that the speed and efficiency of data transmission are not affected in high-node-count servers, thus meeting the needs of large-scale data processing and high-performance computing.

[0101] In other words, in this embodiment, for storage servers with multiple controllers (e.g., four controllers), from front to back, they are a shared module, a computing module, a dedicated module, a management module, and a switching module. The computing module and dedicated module of storage servers with fewer controllers (e.g., two controllers) mentioned above can be reused. The shared module needs to be shared with more nodes, such as two nodes, four nodes, eight nodes, etc. At the same time, the shared device can be a hard drive or an I / O card. Therefore, the shared module is different from storage servers with fewer controllers and needs to be designed separately.

[0102] Furthermore, due to the strong scalability of multi-controller type storage servers, a new switching module is needed. The switching module is also part of the shared module and is shared with multiple nodes. The switching module is located in the front window of the chassis. Through this switching module, more I / O cards or hard drives can be added. Different switching modules can also be replaced according to different expansion needs.

[0103] Furthermore, since the number of storage server nodes in the multi-controller type is large, centralized management is required. Therefore, there is a management module that manages four or two nodes. The management module is also a shared module that is shared with the four or two nodes.

[0104] For example, regarding the connection method between the computing module and the shared module in a multi-controller type storage server, you can refer to... Figure 3 Specifically, such as Figure 3The storage server shown includes: four computing modules, namely computing module 301, computing module 302, computing module 303, and computing module 304; a central backplane 305; two switching modules 306 and 307; and two storage devices 308 and 309. Switching module 306 is configured with two switching expansion cards, switching expansion card 306-1 and switching expansion card 306-2; switching module 307 is configured with two switching expansion cards, switching expansion card 307-1 and switching expansion card 307-2; computing module 301 includes central processing units 301-1 and 301-2; computing module 302 includes central processing units 302-1 and 302-2; computing module 303 includes central processing units 303-1 and 303-2; and computing module 304 includes central processing units 304-1 and 304-2. Furthermore, the central processing unit in each computing module is connected to a switching expansion card in the switching module via a backplane, and the switching expansion card in each switching module is connected to one of the two storage devices. It should be noted that... Figure 3 This is an optional example of the connection method between computing modules and shared modules in a multi-controller type storage server. Specifically, each central processing unit (CPU) in different computing modules may, but is not limited to, be connected to at least one switch expansion card in different switch modules. Each switch expansion card in each switch module is connected to at least one of two storage devices, with the number of CPUs connected to different switch expansion cards evenly distributed, and the number of switch expansion cards connected to different storage devices evenly distributed.

[0105] In other words, multi-controller type storage servers can, but are not limited to, use dual-port Single Root I / O Virtualization (SR-IOV) functionality with NVMe SSDs, coupled with the assistance of a PCIe SwitchFabric architecture, achieving the goals of high performance and efficient resource utilization. Specifically:

[0106] 1) NVMe SSD's Dual-Port SR-IOV Feature: NVMe SSDs, through their dual-port design, support SR-IOV (Single Root I / O Virtualization), which allows a physical storage device to be virtualized into multiple independent virtual functions (VFs). This enables a single physical device to be divided into multiple logical units, each of which can be directly bound to different host nodes, thus achieving flexible resource allocation and efficient concurrent access.

[0107] 2) Introduction of PCIe Switch Fabric mode: By adopting PCIe Switch Fabric mode, the storage system can allocate independent VF resources to each node, avoiding conflicts that may occur when sharing resources, ensuring that each node can have its own dedicated virtual storage space, and greatly improving data read and write speed and system response time.

[0108] 3) Intelligent resource allocation of the node's Basic Input / Output System (BIOS): The node's BIOS is designed to intelligently allocate PCIe bus (Peripheral Component Interconnect Express Bus) and address space resources. It dynamically adjusts the resource allocation to VF according to system requirements, ensuring that each node can operate in the best condition, thereby improving the utilization efficiency of storage devices and the overall performance of the system.

[0109] 4) Standardized NVMe driver access: The NVMe standard driver is responsible for identifying storage devices and connecting them to the system. Through standardized drivers, the compatibility and stability of storage devices are ensured, the device management and maintenance process is simplified, and the integration of devices becomes smoother and more seamless.

[0110] In addition, the system achieves shared communication of the four-controller cache through advanced network links, which not only strengthens the cooperation between nodes, but also ensures real-time synchronization and consistency of data, which is crucial for building a highly available and high-performance storage system.

[0111] For scenarios involving shared I / O cards, there are two main implementation methods: one is to develop a custom multi-host function card, which allows four controller nodes to directly access the same I / O card device, meeting the special requirements for high-concurrency access; the other is to use a standard I / O card in conjunction with a PCIe switching module, and realize data sharing between multiple nodes through a PCIe switching chip (Peripheral Component Interconnect Express Switch, or PCIe switch for short). This method is more flexible, applicable to most standard I / O cards, reduces development costs, and maintains high efficiency in data access.

[0112] To seamlessly integrate with the aforementioned shared module design, a switching module is employed in this embodiment, with the core of this module being the design of the switching chip. The switching chip connects upwards to the CPUs of each controller via high-speed links, and downwards directly to the shared storage devices and I / O cards, forming a clear communication path. The advantage of this design is that the configuration of the switching chip can be flexibly adjusted according to different application scenarios and device types without altering the computing module itself, greatly simplifying the system architecture and improving the adaptability and scalability of the modular design.

[0113] Alternatively, as an alternative example, the module breakdown diagram of a storage server with fewer controller types can, but is not limited to, as shown below. Figure 4 As shown, it includes:

[0114] Shared module 401: Used to provide public storage resources, internally housing hard drive cages and NVMe SSDs.

[0115] Computing module 402: Vertically mounted, with built-in dual CPUs and memory, forming the core of the system's computing.

[0116] Dedicated IO Module 403: Rear-mounted independent card, dedicated to single-node external interface expansion.

[0117] Chassis 404: Provides guide rails, locking and heat dissipation structure, supports cableless plugging and fixing of the above modules, wherein the central backplane can be, but is not limited to, configured between the computing module and the shared module.

[0118] Alternatively, as another alternative example, the module breakdown diagram of a multi-controller type storage server can be, but is not limited to, as shown below. Figure 5 As shown, it includes:

[0119] Shared module 501: Provides public storage resources, and internally houses hard disk cage 501-1 and hard disk 501-2.

[0120] The switching board 502 is horizontally positioned and is responsible for PCIe signal switching and distribution, enabling multi-controller convergence and sharing.

[0121] The computing module 503 is vertically mounted and has built-in dual CPUs and memory, forming the core of the system's computing operations.

[0122] The dedicated IO module 504 is a rear-mounted independent card designed for single-node external interface expansion.

[0123] The 505 centrally located backplane is vertically positioned and integrates high-speed connectors to unify the signal and power interconnection between all modules.

[0124] The frame 506 provides guide rails, locking and heat dissipation structures, and supports cableless plugging and fixing of the above modules.

[0125] Furthermore, as an optional example, the combination relationships of the system modules under different configurations can be, but are not limited to, those shown in Table 1:

[0126] First, the technical terms used in Table 1 will be explained:

[0127] 1) Non-Volatile Memory Express (NVME): This is a high-speed storage interface protocol used to connect solid-state drives (SSDs) to computer systems. It supports PCIe channels and provides high-bandwidth and low-latency data transmission.

[0128] 2) Serial Attached SCSI (SAS): This is a high-speed serial interface used to connect storage devices such as hard drives and tape drives, supporting high data transfer rates and long-distance connections.

[0129] 3) Not Applicable (NA): This means that the component or function at this location is not applicable or used in this configuration.

[0130] 4) Remote Direct Memory Access over Converged Ethernet (ROCE): A network protocol that allows RDMA (Remote Direct Memory Access) traffic to be transmitted over Ethernet for efficient data transmission. It is commonly used in data centers and high-performance computing environments.

[0131] 5) Non-Transparent Bridge (NTB): Used for data transfer between different PCIe domains, often used for resource sharing in multi-node systems.

[0132] 6) Open Compute Project (OCP) Card: A network interface card that conforms to the OCP standard.

[0133] 7) U.2 interface: An interface standard for connecting NVMe solid-state drives, supporting high-speed data transmission, and commonly used in data centers and high-performance storage systems.

[0134] 8) PCI Express Gen 3 (E3): PCIe 3.0 standard, supports higher data transfer rates and is one of the most widely used interface standards.

[0135] 9) 2-controller and 4-controller: The system consists of 2 computing boards (nodes) and belongs to the dual-controller architecture; the system consists of 4 computing boards (nodes) and belongs to the quad-controller architecture.

[0136] 10) 12-drive and 25-drive: The back panel provides 12 hard drive bays, which can accommodate 12 hard drives; or the back panel provides 25 hard drive bays, which can accommodate 25 hard drives.

[0137] 11) 4U: The chassis height is 4 rack units, meaning that the equipment occupies 4 units of space in a 19-inch rack.

[0138] Table 1

[0139]

[0140] In this embodiment of the invention, the computing module is connected to the switching expansion card via a central backplane, and the switching expansion card is connected to the storage device via a high-speed serial bus, forming a high-speed data exchange path. This improves the data transmission speed and the overall system response time, making it suitable for high-performance computing and large-scale data processing scenarios that require high-concurrency data exchange.

[0141] As an optional embodiment, the shared module further includes a shared input / output card for providing input / output services to at least one computing module.

[0142] It should be noted that the aforementioned input / output services may include, but are not limited to, data input acquisition, output transmission, and related control functions. Input / output services are essential in storage servers, ensuring the smooth flow of data and supporting the completion of data processing tasks. Shared input / output cards are designed to enhance these services.

[0143] Optionally, the aforementioned shared input / output card may, but is not limited to, refer to a special type of input / output card that can be shared by multiple computing modules to provide enhanced input / output services. In the shared modules of the storage server, the shared input / output card plays a role in connecting external devices, supporting high-bandwidth data transmission, and optimizing I / O performance, enabling any computing node to efficiently and conveniently access and utilize system peripheral resources, regardless of node location or configuration.

[0144] By adopting the embodiments of this utility model, the shared input / output card included in the shared module allows multiple computing modules to share I / O resources, reducing redundant hardware investment, improving resource utilization efficiency, reducing total cost of ownership, and improving the system's I / O processing capability and response speed.

[0145] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0146] The present invention provides a detailed description of a motherboard. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of these embodiments are only intended to aid in understanding the method and core ideas of the present invention. It should be noted that those skilled in the art can make various improvements and modifications to the present invention without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims

1. A storage server, characterized in that, The storage server includes: at least one computing module, a central backplane, and a shared module, wherein each computing module includes a motherboard, the motherboard includes a central processing unit and memory, the shared module includes a storage device for providing services to the at least one computing module, and each computing module is connected to the shared module through the central backplane.

2. The storage server according to claim 1, characterized in that, The storage server further includes at least one dedicated module, each dedicated module is used to provide services to its corresponding computing module, and each computing module is connected to its corresponding dedicated module.

3. The storage server according to claim 2, characterized in that, Each of the computing modules is connected to its corresponding dedicated module via a high-speed connector.

4. The storage server according to claim 3, characterized in that, Each of the dedicated modules includes at least one of the following: A dedicated management module for managing the computing modules corresponding to each of the dedicated modules; An input / output card used to provide input / output services to the computing module corresponding to each of the dedicated modules; High-speed computer expansion device used to provide extended services to the computing modules corresponding to each of the dedicated modules.

5. The storage server according to claim 1, characterized in that, When the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the central backplane through its corresponding first integrated circuit interconnect bus. The central backplane is connected to the storage device through a second integrated circuit interconnect bus. The first connection channel established between the central processing unit in each computing module and the storage device through the central backplane is the management channel used by each computing module to manage the storage device.

6. The storage server according to claim 5, characterized in that, When the number of computing modules in the storage server is less than a first threshold, the central processing unit in each computing module is connected to the storage device via a high-speed connector, and the second connection channel between the central processing unit in each computing module and the storage device is a data transmission channel used by the storage device to provide data storage services for each computing module.

7. The storage server according to claim 5, characterized in that, If the number of computing modules in the storage server is greater than a second threshold, the shared module further includes: a management module for managing the at least one computing module; and a switching module, wherein the switching module includes at least one switching expansion card, each of the switching expansion cards being configured with at least one switch chip, and the second threshold is greater than or equal to the first threshold.

8. The storage server according to claim 7, characterized in that, When the number of computing modules in the storage server is greater than the second threshold, the central processing unit in each computing module is connected to the switching module through the central backplane, and the switching module is connected to each storage device in the storage device.

9. The storage server according to claim 8, characterized in that, The central processing unit in each computing module is connected to at least one of the switching expansion cards via the central backplane, and the at least one of the switching expansion cards is connected to each of the storage devices in the storage device via a high-speed serial computer expansion bus.

10. The storage server according to any one of claims 1 to 9, characterized in that, The shared module further includes a shared input / output card for providing input / output services to the at least one computing module.

Images