NVMeoF RAID implementation for executing read commands
The NVMeoF RAID controller establishes a virtual connection queue pair between the computing host and the storage target device, solving the scalability and efficiency of the existing storage system, achieving efficient direct data transmission and reducing energy consumption.
Patent Information
- Application Number
- CN201910129096.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-06-15
- Filing Date
- 2019-02-21
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2039-02-21
AI Technical Summary
The existing storage systems rely on PCIe expansion to have scalability, power and latency problems and require complex and expensive RAID hardware, resulting in huge system size and low data transmission efficiency.
Using NVMeoF independent disk redundant array (RAID) controller, the NVMeoF RAID target module, RAID implementation module and RAID initiator module are used to establish a virtual connection queue pair between the computing host and the storage target device to realize direct data transmission, avoiding Latency and power consumption caused by multiple DMA data transactions.
It realizes efficient direct data transmission between the computing host and the storage target device, reduces the system's volume and energy consumption, improves scalability, and reduces dependence on RAID hardware.
Smart Images

Figure CN110609659B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates generally to storage systems and, more particularly, to systems and methods for performing efficient RAID implementation of read commands via virtual connection queue pairs in NVMeoF Ethernet-based SSDs. Background Art
[0002] RAID (Redundant Array of Independent Disks) is a storage technology that combines multiple disk drive assemblies into logical units. Data is distributed across the drives in one of a number of ways, known as "RAID levels," depending on the level of redundancy and performance. A solid-state drive (SSD) or solid-state disk is a solid-state storage device that uses an integrated circuit assembly as memory to store data persistently. SSD technology primarily uses an electronic interface that is compatible with traditional block input / output (I / O) hard disk drives (HDDs), which allows for simple replacement in common applications.
[0003] Existing storage systems are typically physical servers (storage box nodes) with one or more hard disk drives (HDDs) or one or more solid-state drives (SSDs). Existing storage systems may also be virtual machines or virtual logic unit components that can access one or more HDDs and / or SSDs. Existing storage systems may have PCIe (Peripheral Component Interconnect Express) expansion limitations. As shown in Figure 1, existing storage systems have a large volume because they require CPUs, expensive complex motherboards with RAID hardware. In addition, existing storage systems use multiple direct memory access (DMA) data copies to perform data transfers. These multiple DMA data transactions cause more delays and consume more power.
[0004] FIG. 1 is a schematic diagram showing an existing architecture with a storage box node 106 for data transmission according to the prior art. According to the prior art, one or more computer hosts 102_1 to 102_M are connected to a 40G / 100G Ethernet switch 104, which is in turn connected to the storage box node 106. The storage box node 106 includes one or more SSDs 108_1 to 108_N with RAID hardware and has PCIe expansion limitations. The storage box node 106 also includes a CPU, a motherboard with RAID hardware, which is expensive and makes the storage box node 106 bulky. In addition, data transmission between the storage box node 106 and one or more computing hosts 102_1 to 102_M is performed using multiple data transactions, which create data traffic, consume power, and increase latency.
[0005] Therefore, there is a need for a system and method that solves the above problems and difficulties in this article and provides solutions. Therefore, there is still a need for a system that solves the problems of scalability (Ethernet SSD scalability methods other than PCIe), power and latency, and opens up the possibility of directly transmitting data between storage nodes and computing hosts. Summary of the invention
[0006] Various embodiments of the present invention describe an NVMeoF redundant array of independent disks (RAID) controller for performing direct data transfer between a computing host and one or more storage target devices based on a structure. The NVMeoF redundant array of independent disks (RAID) controller includes an NVMeoF RAID target module, an NVMeoF RAID implementation module, and an NVMeoF RAID initiator module. The NVMeoF RAID target module is suitable for interacting with the computing host and receiving one or more non-volatile memory express over fabrics (NVMeoF) commands based on a structure from the computing host. The NVMeoF RAID implementation module is suitable for receiving the one or more NVMeoF commands from the NVMeoF RAID target module and performing RAID functions on the received one or more NVMeoF commands. The NVMeoF RAID initiator module is suitable for interacting with the one or more storage target devices. The NVMeoF RAID initiator module receives the one or more NVMeoF commands from the NVMeoF RAID implementation module, transmits the one or more NVMeoF commands to the one or more storage target devices, and establishes one or more virtual queue pair connections between the computing host and the one or more storage target devices to implement the direct data transmission between the computing host and the one or more storage target devices.
[0007] According to one embodiment, the direct data transmission between the computing host and the one or more storage target devices is realized through the following connections: a computing-side physical connection established by the computing host between the computing host and the NVMeoF RAID target module, a storage-side physical connection established by the NVMeoF RAID initiator module between the NVMeoF RAID initiator module and the one or more storage target devices, and one or more virtual queue pair connections established by the NVMeoF RAID initiator module between the one or more storage target devices and the computing host.
[0008] According to one embodiment, the RAID functionality includes identifying one or more corresponding storage target devices, wherein the NVMeoF command for the one or more corresponding storage target devices is transmitted by the computing host.
[0009] In one aspect, a method for establishing a computing-side physical connection between a computing host and an NVMeoF RAID target module is provided, the method comprising: issuing, by the computing host, one or more NVMeoF connection commands to the NVMeoF RAID target module to establish the computing-side physical connection; receiving, by the NVMeoF RAID target module, the one or more NVMeoF connection commands issued from the computing host to create management and input / output (I / O) queues within the NVMeoF RAID target module; transmitting, by the NVMeoF RAID target module, one or more NVMeoF connection command completions to the computing host; and receiving, by the computing host, the one or more NVMeoF connection command completions from the NVMeoF RAID target module when the computing-side physical connection is established.
[0010] On the other hand, a method for establishing one or more storage-end physical connections between an NVMeoF RAID initiator module and one or more storage target devices is provided, the method comprising: issuing one or more NVMeoF connection commands to the one or more storage target devices by the NVMeoF RAID initiator module to establish the one or more storage-end physical connections; receiving the one or more NVMeoF connection commands from the NVMeoF RAID initiator module by the one or more storage target devices to create, manage and implement one or more I / O queues in the one or more storage target devices; transmitting one or more NVMeoF connection command completions to the NVMeoF RAID initiator module by the one or more storage target devices; and receiving the one or more NVMeoF connection command completions from the one or more storage target devices by the NVMeoF RAID initiator module when the one or more storage-end physical connections are established.
[0011] On the other hand, a method for establishing one or more virtual queue pair connections is provided, wherein the one or more virtual queue pair connections are used to realize direct data transmission between one or more computing hosts and one or more storage target devices, the method comprising: issuing one or more vendor connection commands to the one or more storage target devices by an NVMeoF RAID initiator module, wherein the one or more vendor connection commands include information about one or more computing hosts used to establish the one or more virtual queue pair connections between the one or more computing hosts and the one or more storage target devices; receiving the one or more vendor connection commands by the one or more storage target devices to create virtual management and realize I / O queue connection between the one or more computing hosts and the one or more storage target devices; when the one or more virtual queue pair connections are established, transmitting the one or more vendor connection command completions to the NVMeoF RAID initiator module by the one or more storage target devices; and receiving the one or more vendor connection command completions from the one or more storage target devices by the NVMeoF RAID initiator module.
[0012] According to one embodiment, the one or more virtual queue pair connections are established by creating one or more compute-side physical connections between the one or more computing hosts and the NVMeoF RAID controller and creating one or more storage-side physical connections between the NVMeoF RAID controller and the one or more storage target devices.
[0013] In another aspect, a method for reading input / output (I / O) data between a computing host and one or more storage target devices is provided, the method comprising: sending, by the computing host, one or more NVMeoF read commands to an NVMeoF RAID target module for reading data in n logical blocks (nLB) of the one or more storage target devices starting at a start logical block address (sLBA); identifying, by an NVMeoF RAID implementation module, the one or more storage target devices storing data from the nLB; initiating, by an NVMeoF RAID initiator module, the one or more NVMeoF read commands to the identified one or more storage target devices including data from corresponding sLBAs and nLBs; initiating, by the one or more storage target devices, RDMA_WRITE transactions directly to the computing host using one or more virtual queues; transmitting, by the one or more storage target devices, one or more successful read command completions to the NVMeoF RAID initiator module; and transmitting, by the NVMeoF RAID target module, the one or more successful read command completions from the one or more storage target devices to the computing host as a merged single command completion.
[0014] In another aspect, a method for performing input / output (I / O) read fault-tolerant data between a computing host and one or more storage target devices is described herein. The method includes: transmitting, by the computing host, one or more NVMeoF read commands to an NVMeoF RAID target module for reading data from the one or more storage target devices of nLB starting from an sLBA; identifying, by the NVMeoF RAID controller module, the one or more storage target devices storing data from the nLB starting from the sLBA; initiating, by the NVMeoF RAID initiator module, the one or more NVMeoF read commands to the identified one or more registered storage target devices including data from corresponding sLBAs and nLBs based on one or more storage-side physical connections; initiating, by the one or more storage target devices, one or more RDMA_WRITE transactions directly to the computing host by using one or more virtual queue pair connections; transmitting, by the one or more storage target devices, one or more read command completions to the NVMeoF RAID initiator module by using the one or more storage-side physical connections; obtaining, by the NVMeoF RAID initiator module, a read completion failure and identifying a cause based on the one or more storage-side physical connections to reconstruct data for the read completion failure; The RAID initiator module reissues the one or more NVMeoF read commands to the one or more registered storage target devices based on the one or more storage-side physical connections; the one or more storage target devices transmit the corresponding stripes and parity stripes to the NVMeoF RAID initiator module through the one or more RDMA_WRITE transactions based on the one or more storage-side physical connections; the NVMeoF RAID controller retrieves the missing stripes from the received stripes and parity stripes; the one or more storage target devices transmit one or more successful NVMeoF read completions to the NVMeoF RAID initiator module; the NVMeoF RAID target module transmits the retrieved stripes as RDMA_WRITE transactions to the computing host; and the NVMeoF RAID target module of the NVMeoF RAID controller sends a merged NVMeoF command completion to the computing host.
[0015] The above has been summarized various aspects of the present invention, and is used to help better understand the subsequent more complete detailed description. With reference to these, it should be clearly understood that the present invention is not limited to the methods or applications described and illustrated herein. Any other advantages and purposes of the present invention that are obvious or apparent from the detailed description or description contained herein are within the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] According to the following description of the preferred embodiments and the accompanying drawings, those skilled in the art will appreciate other purposes, features and advantages, among which:
[0017] FIG. 1 is a schematic diagram showing a conventional architecture having a storage box node for data transmission according to the prior art.
[0018] Figure 2 is a schematic diagram illustrating an exemplary method for efficient RAID implementation through virtual connection queue pairs in a storage target device according to an embodiment of the present invention.
[0019] Figure 3 is a schematic flow chart showing a method for establishing one or more virtual queue pair connections between a computing host and one or more storage target devices using an NVMeoF RAID controller according to an embodiment of the present invention.
[0020] Figure 4 is a schematic flow chart illustrating reading an input / output (I / O) data stream between a computing host and one or more storage target devices according to an embodiment of the present invention.
[0021] Figure 5 The present invention is a schematic flow chart showing a method of performing input / output (I / O) reading fault-tolerant data between a computing host and one or more storage target devices according to an embodiment of the present invention.
[0022] Although specific features of the invention are shown in some drawings and not in others, this is for convenience only as each feature may be combined with any or all of the other features in accordance with the invention. DETAILED DESCRIPTION
[0023] The present invention describes a system and method for executing an efficient RAID implementation of a read command through a virtual connection queue pair in an Ethernet-based storage target device. In the following detailed description of embodiments of the present invention, reference is made to the accompanying drawings that form a part of this document, and the accompanying drawings illustrate specific embodiments in which the present invention can be implemented. These embodiments are described in sufficient detail to enable those skilled in the art to implement the present invention, and it should be understood that other embodiments can be utilized and changes can be made without departing from the scope of the present invention. Therefore, the following detailed description should not be regarded as having a limiting meaning, and the scope of the present invention is limited only by the appended claims.
[0024] The specification may refer to "one", "an" or "some" embodiments in several places. This does not necessarily mean that each such reference is to the same embodiment, or that the feature applies to only a single embodiment. Individual features of different embodiments may also be combined to provide other embodiments.
[0025] As used herein, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless expressly stated otherwise. It will be further understood that when the terms "comprises", "comprising", "including", and / or "comprising" are used in this specification, it refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups of the specification. As used herein, the term "and / or" includes any and all combinations and arrangements of one or more of the relevant listed items.
[0026] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those commonly understood by those skilled in the art to which the present disclosure belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and will not be understood as an idealized or overly formal meaning unless explicitly defined in this article.
[0027] By referring to the non-limiting embodiments shown in the drawings and described in detail in the following description, the embodiments of this article and their various features and advantages will be explained more fully. Descriptions of well-known components and processing techniques have been omitted so as not to unnecessarily obscure the embodiments of this article. The examples used herein are only intended to facilitate understanding of the manner in which the embodiments herein can be implemented, and further enable those skilled in the art to implement the embodiments of this article. Therefore, these embodiments should not be interpreted as limiting the scope of the embodiments of this article.
[0028] In the following detailed description of embodiments of the present invention, reference is made to the accompanying drawings which form a part of this document, and in which specific embodiments in which the present invention may be implemented are shown by way of illustration. These embodiments are described in sufficient detail to enable those skilled in the art to implement the present invention, and it should be understood that other embodiments may be utilized and may be changed without departing from the scope of the present invention. Therefore, the following detailed description should not be regarded as having a limiting meaning, and the scope of the present invention is limited only by the appended claims.
[0029] Figure 22 is a schematic diagram showing an exemplary method for efficient RAID implementation through virtual connection queue pairs in storage target devices 208_1 to 208_N according to an embodiment of the present invention. According to the present embodiment, data transmission between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N is performed directly using virtual connection queue pairs created by the NVMeoF RAID controller 204. The NVMeoF RAID controller 204, one or more computing hosts 202_1 to 202_M, and one or more storage target devices 208_1 to 208_N are connected to an Ethernet switch 206. The NVMeoF RAID controller 204 includes an NVMeoF RAID target module 210, an NVMeoF RAID implementation module 212, and an NVMeoF RAID initiator module 214. The NVMeoF RAID target module 210 is adapted to interact with one or more computing hosts 202_1 to 202_M and receive one or more fabric-based high-speed non-volatile memory (NVMeoF) commands from the one or more computing hosts 202_1 to 202_M. The NVMeoF RAID implementation module 212 is adapted to receive one or more NVMeoF commands from the NVMeoF RAID target module 210 and perform RAID functions on the one or more received NVMeoF commands.
[0030] The NVMeoF RAID initiator module 214 receives one or more NVMeoF commands from the NVMeoF RAID implementation module 212, and transmits the one or more NVMeoF commands to one or more storage target devices 208_1 to 208_N. The NVMeoF RAID initiator module 214 establishes one or more virtual queue pair connections between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N, thereby enabling direct data transmission between the one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N. Direct data transmission between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N is achieved through the following connections: one or more computing-side physical connections established between one or more computing hosts 202_1 to 202_M and the NVMeoF RAID target module 210, one or more storage-side physical connections established between the NVMeoF RAID initiator module 214 and one or more storage target devices 208_1 to 208_N, and one or more virtual queue pair connections between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N.
[0031] Figure 3 2 is a schematic flow chart showing a method for establishing one or more virtual queue pair connections between a computing host 202_1 and one or more storage target devices 208_1 to 208_N using an NVMeoF RAID controller 204 according to an embodiment of the present invention. The connection between the computing host 202_1 and the Ethernet 206, the connection between the NVMeoF RAID controller 204 and the Ethernet 206, and the connection between the storage target devices 208_1 to 208_N and the Ethernet 206 are not shown in the figure. According to the figure, at step 0, the computing host 202_1 performs exploration and discovers the NVMeoF RAID controller 204. At step 1, the computing host 202_1 sends an NVMeoF connection command to the NVMeoF RAID controller 204. At step 2, the NVMeoF RAID target module 210 sends the completion of the NVMeoF connection command to the computing host 202_1. At step 3, the computing host 202_1 establishes a computing-side physical connection with the NVMeoF RAID controller 204. The NVMeoF RAID implementation module 212 provides the information of the registered storage target to the NVMeoF RAID initiator module 214. At step 4 (i.e., 4a-4n), the NVMeoF RAID initiator module 214 sends (a) one or more NVMeoF connection commands with the transfer context information of the NVMeoF RAID initiator module 214 and (b) one or more vendor connection commands with the transfer context information of the computing host 202_1 to the registered one or more storage target devices 208_1 to 208_N to establish one or more storage-side physical connections. In an embodiment, each of the one or more vendor connection commands transmitted to the registered one or more storage target devices 208_1 to 208_N by the NVMeoF RAID initiator module 214 is a different / unique command from each other.
[0032] At step 5 (i.e., 5a-5n), after the internal storage NVMeoF RAID initiator module 214 and the computing host 202_1 transmit context information, the one or more storage target devices 208_1 to 208_N send the NVMeoF connection command completion and the vendor connection command completion to the NVMeoF RAID initiator module 214. At step 6 (i.e., 6a-6n), the NVMeoF RAID initiator module 214 establishes one or more storage-side physical connections with the registered one or more storage target devices 208_1 to 208_N. At step 7 (i.e., 7a-7n), the one or more storage target devices 208_1 to 208_N establish one or more virtual queue pair connections with the computing host 202_1. The one or more virtual queue pair connections allow direct data transfer between the one or more storage target devices 208_1 to 208_N and the computing host 202_1. In an embodiment, one or more virtual queue pair connections are established between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N. The one or more virtual queue pair connections are established through one or more computing-side physical connections, one or more storage-side physical connections, and by storing the transfer context information of the computing hosts 202_1 to 202_M and the transfer context information of the NVMeoF RAID initiator module 214 in the one or more storage target devices 208_1 to 208_N.
[0033] Figure 42 is a schematic flow chart showing the reading of input / output (I / O) data between a computing host 202_1 and one or more storage target devices 208_1 to 208_N according to an embodiment of the present invention. The connection between the computing host 202_1 and the Ethernet 206, the connection between the NVMeoF RAID controller 204 and the Ethernet 206, and the connection between the storage target devices 208_1 to 208_N and the Ethernet 206 are not shown in the figure. According to the figure, at step 1, the computing host 202_1 sends an NVMeoF read command to the NVMeoF RAID target module 210 on the computing end physical connection, for reading data of n logical blocks (nLB) starting from the start logical block address (sLBA). At step 2, the NVMeoF RAID implementation module 212 identifies one or more storage target devices 208_1 to 208_N storing data from nLB. At step 3 (i.e., 3a-3n), the NVMeoF RAID initiator module 214 initiates the following transmission: one or more NVMeoF read commands on the one or more storage-side physical connections are transmitted to the identified one or more storage target devices 208_1 to 208_N (e.g., one or more registered storage target devices) including data from the corresponding sLBA and nLB. At step 4 (i.e., 4a-4n), the one or more storage target devices 208_1 to 208_N directly initiate RDMA_WRITE transactions to the computing host 202_1 by using the established one or more virtual queue pairs to connect. At step 5 (i.e., 5a-5n), the one or more storage target devices 208_1 to 208_N transmit one or more successful read completions to the NVMeoF RAID initiator module 214 on the one or more storage-side physical connections. Finally, at step 6, the NVMeoF RAID target module 210 transmits a consolidated single command completion to the computing host 202_1 on the computing-side physical connection. In an embodiment, the NVMeoF RAID controller 204 is used to execute one or more NVMeoF read commands between one or more computing hosts 202_1 to 202_M and one or more storage target devices 208_1 to 208_N through the above steps.
[0034] Figure 51 is a schematic flow chart showing an input / output (I / O) NVMeoF read fault-tolerant data between a computing host 202_1 and an NVMeoF RAID controller 204 according to an embodiment of the present invention. The connection between the computing hosts 202_1 to 202_M and the Ethernet 206, the connection between the NVMeoF RAID controller 204 and the Ethernet 206, and the connection between the storage target devices 208_1 to 208_N and the Ethernet 206 are not shown in the figure. At step 1, the computing host 202_1 sends an NVMeoF read command to the NVMeoF RAID target module 210 to read data of n logical blocks (nLB) starting from the start logical block address (sLBA). At step 2, the NVMeoF RAID implementation module 212 identifies one or more storage target devices 208_1 to 208_N storing nLB of data. At step 3 (ie, 3a-3n), the NVMeoF RAID initiator module 214 initiates the transmission of one or more NVMeoF read commands on one or more storage terminal physical connections to the identified one or more storage target devices 208_1 to 208_N including data from corresponding sLBAs and nLBs.
[0035] At step 4 (i.e., 4a-4n), one or more storage target devices 208_1 to 208_N directly initiate one or more RDMA_WRITE transactions to the computing host 202_1 by connecting using one or more virtual queue pairs for data transfer of corresponding logical blocks. At step 5 (i.e., 5a-5n), one or more storage target devices 208_1 to 208_N also transmit one or more read command completions to the NVMeoF RAID initiator module 214 on one or more storage-side physical connections. The NVMeoF RAID initiator module 214 obtains a read completion failure command through one of the one or more storage-side physical connections and identifies the cause to rebuild data for the read completion failure.
[0036] At step 6 (i.e., 6a-6n), the NVMeoF RAID initiator module 214 reissues one or more NVMeoF read commands to the registered (previously sent successful completion) one or more storage target devices 208_1 to 208_N. The one or more NVMeoF read commands are reissued to request the one or more storage target devices 208_1 to 208_N to send data to the NVMeoF RAID initiator module 214 based on one or more storage-side physical connections. At step 7 (i.e., 7a-7n), the one or more storage target devices 208_1 to 208_N initiate one or more RDMA_WRITE transactions to the NVMeoF RAID initiator module 214 through one or more storage-side physical connections for data transfer of corresponding logical blocks. At step 8 (i.e., 8a-8n), the one or more storage target devices 208_1 to 208_N transmit one or more successful NVMeoF read completions to the NVMeoF RAID initiator module 214. At step 9, the NVMeoF RAID controller 204 retrieves the missing stripes (for which the NVMeoF RAID initiator module 214 previously received a failed completion) from the received stripes and parity stripes. At step 10, the NVMeoF RAID target module 210 transmits the retrieved stripes as RDMA_WRITE transactions to the computing host 202_1. Finally, at step 11, the NVMeoF RAID target module 210 of the NVMeoF RAID controller 204 sends the merged NVMeoF command completion to the computing host 202_1. In an embodiment, fault tolerance of one or more NVMeoF read commands is processed between one or more computing hosts 202_1 to 202_M and the NVMeoF RAID controller 204 via the above steps.
[0037] Although this specification has been described in detail with respect to specific embodiments of the present invention, it should be understood that those skilled in the art, after understanding the foregoing, can easily think of changes, variations and equivalents to these embodiments. Without departing from the scope of the present invention, those of ordinary skill in the art can implement these and other modifications and variations of the present invention. In addition, those of ordinary skill in the art will understand that the foregoing description is merely exemplary and is not intended to limit the present invention. Therefore, this subject matter is intended to cover these modifications and variations.
Claims
1. A method for establishing a connection between a computer host and a first storage device via an intermediate device, the method include: The computer host issues a first structure-based high-speed non-volatile memory connection command to the structure-based high-speed non-volatile memory independent disk redundant array target module of the intermediate device, so as to establish a computer-side physical connection between the structure-based high-speed non-volatile memory independent disk redundant array target module of the intermediate device and the computer host; In response to receiving the first fabric-based high-speed non-volatile memory connection command from the computer host, the fabric-based high-speed non-volatile memory redundant array of independent disks target module of the intermediate device creates a first input / output queue in the fabric-based high-speed non-volatile memory redundant array of independent disks target module of the intermediate device; The structure-based high-speed non-volatile memory independent disk redundant array target module of the intermediate device completes the transmission of the first structure-based high-speed non-volatile memory connection command to the computer host; The computer host receives the completion of the first fabric-based high-speed non-volatile memory connection command from the fabric-based high-speed non-volatile memory redundant array of independent disks target module of the intermediate device; The structure-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device issues a second structure-based high-speed non-volatile memory connection command to the first storage device to establish a first storage-end physical connection between the first storage device and the structure-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device; In response to receiving the second fabric-based high-speed non-volatile memory connection command from the fabric-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device, the first storage device creates a second input / output queue in the first storage device; The first storage device transmits the second structure-based high-speed non-volatile memory connection command to the structure-based high-speed non-volatile memory independent disk redundant array initiator module of the intermediate device; The fabric-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device receives the second fabric-based high-speed non-volatile memory connection command completion from the first storage device; and First data for direct memory access is transmitted from the first storage device to the computer host using the computer-side physical connection and the first storage-side physical connection through the first input / output queue and the second input / output queue.
2. The method according to claim 1, further comprising: include: The structure-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device issues a third structure-based high-speed non-volatile memory connection command to the second storage device to establish a second storage-end physical connection between the second storage device and the structure-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device; In response to receiving the third fabric-based high-speed non-volatile memory connection command from the fabric-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device, the second storage device creates a third input / output queue in the second storage device; The second storage device transmits the third structure-based high-speed non-volatile memory connection command to the structure-based high-speed non-volatile memory independent disk redundant array initiator module of the intermediate device; The fabric-based high-speed non-volatile memory redundant array of independent disks initiator module of the intermediate device receives the third fabric-based high-speed non-volatile memory connection command completion from the second storage device; and The first data of the direct memory access is transmitted from the first storage device and the second storage device to the computer host using the computer-side physical connection, the first storage-side physical connection and the second storage-side physical connection through the first input / output queue, the second input / output queue and the third input / output queue.
Citation Information
Patent Citations
Redundant array of independent disks (raid) write cache sub-assembly
CN102203751A
Techniques to Replicate Data between Storage Servers
CN104283938A
