CN119071310A - Cloud storage processing method, device, storage medium and system - Google Patents

Abstract

The application provides a cloud storage processing method, equipment, a storage medium and a system, wherein the method comprises the steps that a special processing chip configured to provide a data surface function in a computing server receives a data read-write request sent by a client virtual machine, the data read-write request comprises a logic disk storage address corresponding to data to be read-written, a read-write request control signal containing the logic disk storage address is sent to a general processing chip configured to provide a control surface function in the computing server, so that the general processing chip generates header information, the header information comprises identification information of a target storage server corresponding to the logic disk storage address, a message sending control signal comprising the header information is received, and a network request message is generated according to the header information and the data read-write request and is sent to the target storage server. The architecture that the data surface and the control surface are separated is adopted, so that the processing delay of the data read-write request is reduced, and the processing throughput is improved.

Description

Cloud storage processing method, device, storage medium and system

Technical Field

The present invention relates to the field of cloud computing technologies, and in particular, to a cloud storage processing method, device, storage medium, and system.

Background

With the continuous evolution of storage technology, storage media range from traditional mechanical hard disks (HARD DISK DRIVE, HDD for short), solid state disks (Solid STATE DRIVE, SSD for short), and to recent aoteng (Optane) SSD. The write latency of the storage medium decreases from the millisecond level to the microsecond level and the throughput increases from the MB/s level to the GB/s level. Meanwhile, network bandwidth of data centers of cloud manufacturers is also continuously improved.

Cloud storage systems such as cloud disks are sensitive to data read-write latency, and low-latency high-throughput cloud storage systems are a design goal. At present, a cloud storage system mostly adopts a framework with separated computation and storage, and generally comprises a computing server cluster and a storage server cluster, wherein the storage server cluster provides a physical storage space for data storage, one or more virtual machines can be deployed in each computing server, the virtual machines respectively correspond to different users, application programs of the corresponding users are operated in the computing server, in addition, a cloud storage service is operated in the computing server, and the virtual machines trigger data read-write requests to the storage server cluster through the cloud storage service.

In a traditional cloud storage system, a client virtual machine generates a data read-write request and then sends the data read-write request to a CPU in a computing server, the CPU processes related data, and based on a network protocol type between a storage server cluster and the computing server, a control strategy is transmitted through a plurality of networks to generate a network message sent to a certain storage server, and the network message is sent to the storage server through a network card, so that the storage server completes corresponding data read-write operation. The data processing process executed by the CPU often needs to go through the processes of a virtualized software stack, a storage software stack, a network protocol stack, and the like, and multiple related data copying and converting operations (such as reading data from a memory or writing data into the memory, copying the data to a network card, performing security processing such as encryption and decryption on the data, and the like) can occur in the middle, so that the processing time is long.

In practice, the computing server often includes multiple CPUs, and the data processing process and the network transmission process need to take up several CPUs to complete, which means that enough CPUs need to be reserved to complete the processing processes, and the remaining CPUs can only be used to construct the client virtual machine, so that the CPU resources cannot be utilized better. Moreover, the CPU takes longer time to complete the data processing process and the network transmission process (such as copying the data on the memory, the network card, and packaging the network packet) as a whole, which results in greater processing delay of the data read-write request and finally results in poor throughput performance.

Disclosure of Invention

The embodiment of the invention provides a cloud storage processing method, equipment, a storage medium and a system, which are used for reducing the data read-write processing delay of cloud storage and improving the throughput performance.

In a first aspect, an embodiment of the present invention provides a cloud storage processing method, applied to a dedicated processing chip configured to provide a data plane function in a computing server, where the method includes:

receiving a data read-write request sent by a client virtual machine, wherein the data read-write request comprises a logic disk storage address corresponding to data to be read-written;

Transmitting a read-write request control signal containing the logical disk storage address to a general processing chip configured to provide a control plane function in the computing server, so that the general processing chip generates header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address;

receiving a message sending control signal sent by the general processing chip, wherein the message sending control signal comprises the header information;

Generating a network request message according to the header information and the data read-write request, and sending the network request message to the target storage server so that the target storage server processes the data read-write request.

In a second aspect, an embodiment of the present invention provides a cloud storage processing apparatus applied to a dedicated processing chip configured to provide a data plane function in a computing server, the apparatus including:

The receiving module is used for receiving a data read-write request sent by the client virtual machine, wherein the data read-write request comprises a logic disk storage address corresponding to data to be read-written;

a sending module, configured to send a read-write request control signal containing the logical disk storage address to a general processing chip configured to provide a control plane function in the computing server, so that the general processing chip generates header information, where the header information includes identification information of a target storage server corresponding to the logical disk storage address;

the receiving module is further configured to receive a message sending control signal sent by the general processing chip, where the message sending control signal includes the header information;

the generation module is used for generating a network request message according to the header information and the data read-write request;

The sending module is further configured to send the network request packet to the target storage server, so that the target storage server processes the data read-write request.

In a third aspect, an embodiment of the present invention provides a cloud storage processing method, applied to a general processing chip configured to provide a control plane function in a computing server, where the method includes:

Receiving a read-write request control signal sent by a special processing chip configured to provide a data surface function in the computing server, wherein the read-write request control signal is sent after the special processing chip receives a data read-write request sent by a client virtual machine, the data read-write request comprises a logic disk storage address corresponding to data to be read-written, and the read-write request control signal comprises the logic disk storage address;

Generating header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address;

and sending a message sending control signal containing the header information to the special processing chip so that the special processing chip generates a network request message according to the header information and the data read-write request, and sending the network request message to the target storage server so that the target storage server processes the data read-write request.

In a fourth aspect, an embodiment of the present invention provides a cloud storage processing apparatus applied to a general purpose processing chip configured to provide a control plane function in a computing server, the apparatus including:

The receiving module is used for receiving a read-write request control signal sent by a special processing chip configured to provide a data surface function in the computing server, wherein the read-write request control signal is sent after the special processing chip receives a data read-write request sent by a client virtual machine, the data read-write request comprises a logic disk storage address corresponding to data to be read-written, and the read-write request control signal comprises the logic disk storage address;

the generation module is used for generating header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address;

and the sending module is used for sending a message sending control signal containing the header information to the special processing chip so that the special processing chip generates a network request message according to the header information and the data read-write request, and sending the network request message to the target storage server so that the target storage server processes the data read-write request.

In a fifth aspect, an embodiment of the present invention provides a computing server, including a memory, a general-purpose processing chip, a special-purpose processing chip, and a communication interface, where the memory stores executable codes, and when the executable codes are executed by the general-purpose processing chip and the special-purpose processing chip, the cloud storage processing method according to the first aspect or the third aspect is executed.

In a sixth aspect, embodiments of the present invention provide a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a general purpose processing chip and a special purpose processing chip in a computing server, can implement at least the cloud storage processing method according to the first or third aspect.

In a seventh aspect, an embodiment of the present invention provides a cloud storage system, including:

A computing server cluster and a storage server cluster, wherein any computing server in the computing server cluster comprises a client virtual machine, a special processing chip configured to provide a data plane function and a general processing chip configured to provide a control plane function;

the special processing chip is used for executing the cloud storage processing method according to the first aspect;

The general processing chip is used for executing the cloud storage processing method according to the third aspect.

In the cloud storage processing scheme provided by the embodiment of the invention, a distributed cloud storage system is provided, wherein the distributed cloud storage system is formed by a computing server cluster and a storage server cluster, a client virtual machine using cloud storage service, a special processing chip configured to provide a data plane function and a general processing chip configured to provide a control plane function are included in any computing server in the computing server cluster, wherein the special processing chip is a data processor (Data Processing Unit, abbreviated as DPU), and the general processing chip is a CPU. In general, the control plane function is mainly a network transmission control function, such as a check logic including a control logic for sending a data read-write request to a network of a storage server and a network response message fed back by the storage server, and the data plane function is mainly a processing logic of data, such as copying, encrypting, packaging of a network message, and the like. The special processing chip is better at performing data processing work with simple processing logic, is not better than complex network transmission protocol processing work, and the general processing chip is opposite, so that the processing tasks of data read-write requests triggered by the client virtual machine can be completed by combining the advantages of the general processing chip and the special processing chip, better performance can be obtained, a small number of general processing chips can be distributed for the computing server to provide the control surface function, more general processing chips can be used for constructing more client virtual machines, and resources of the general processing chips can be more fully utilized.

Specifically, after receiving a data read-write request sent by a client virtual machine, the special processing chip analyzes the data read-write request, generates a read-write request control signal containing a logic disk storage address corresponding to data to be read-written therein, and sends the read-write request control signal to the general processing chip. The general processing chip generates header information based on the set network transmission strategy, the header information comprises identification information of a target storage server corresponding to the logic disk storage address, and a message sending control signal containing the header information is sent to the special processing chip. The special processing chip encapsulates the data read-write request according to the header information to generate a network request message, and sends the network request message to the target storage server so that the target storage server processes the data read-write request, namely, the corresponding data read-write operation is completed.

Therefore, the network card function is actually integrated on the special processing chip, the network message packaging processing can be directly carried out on the data read-write request based on the header information fed back by the general processing chip, the data conversion operation is reduced (for example, the packaging of the network request message is directly carried out after the data to be written is directly copied from the memory aiming at the data write request, the network card is not required to be copied, and the data copying operation is reduced), thereby the processing delay of the data read-write request is reduced, and the processing throughput is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic diagram of a cloud storage processing system according to an embodiment of the present invention;

Fig. 2 is an interaction flow chart of a cloud storage processing method provided by an embodiment of the present invention;

fig. 3 is an interaction flow chart of a cloud storage processing method according to an embodiment of the present invention;

Fig. 4 is an interaction flow chart of a cloud storage processing method according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a data plane according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a control surface according to an embodiment of the present invention;

fig. 7 is a schematic diagram of a processing procedure of a control plane in a network message sending process according to an embodiment of the present invention;

Fig. 8 is a schematic diagram of a processing procedure of a control plane in a network packet receiving process according to an embodiment of the present invention;

Fig. 9 is a schematic structural diagram of a cloud storage processing device according to an embodiment of the present invention;

Fig. 10 is a schematic structural diagram of a cloud storage processing device according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of a computing server according to the present embodiment.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. In addition, the sequence of steps in the method embodiments described below is only an example and is not strictly limited.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the embodiments of the present invention are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide corresponding operation entries for the user to select authorization or rejection.

Some concepts involved in the embodiments of the present invention will be explained first.

A data processor (Data Processing Unit, abbreviated as DPU) is a broad class of special purpose processors, is another important computational power chip in a data center scenario, following a CPU (Central Processing Unit ), GPU (Graphics Processing Unit, graphics processor), to provide a compute engine for high bandwidth, low latency, data intensive computing scenarios.

The Non-volatile memory host controller interface specification (Non-Volatile Memory Express, NVMe) is that the NVMe specially designs a data access protocol for high-performance equipment, is simpler and lighter, and can greatly improve the storage performance by matching with a multi-queue technology.

Virtio-Virtio is an IO paravirtualization solution that provides a communication framework and programming interface between a guest OS (guest operating system) and a hypervisor (a virtualization manager) virtualization device. The method has the advantages of improving performance and reducing compatibility problems caused by cross-platform, and is widely applied to cloud computing.

Direct memory access (Direct Memory Access, DMA) is a function provided by some computer bus architecture that enables additional devices (e.g., disk drives) to send or read data directly from the memory of a computer motherboard.

Fig. 1 is a schematic diagram of a cloud storage processing system according to an embodiment of the present invention, where, as shown in fig. 1, the cloud storage processing system includes a computing server cluster and a storage server cluster.

The computing server cluster and the storage server cluster may be physically isolated, that is, the cloud storage system provided by the embodiment of the present invention may adopt a architecture in which storage and computation are separated.

Wherein a cluster of storage servers may include several storage servers, each of which provides a physical storage medium, such as various types of disks. Only general purpose processing chips may be deployed in these storage servers for local data access processing.

Wherein a cluster of compute servers may include several compute servers, each including a guest virtual machine, a dedicated processing chip configured to provide data plane functionality (illustrated as an I/O data plane in fig. 1), and a general purpose processing chip configured to provide control plane functionality (illustrated as an I/O control plane in fig. 1). The I/O represents input/output, and corresponds to the access process of the client virtual machine for writing data and reading data respectively according to the data transmission direction. In practice, a cloud storage service, such as a common cloud disk service, is also running in the computing server, and each client virtual machine invokes the cloud storage service to perform data read-write processing on the storage server cluster.

The client virtual machine can be virtualized based on hardware resources such as a general processing chip in the corresponding computing server. At least one client virtual machine can be constructed in one computing server so as to be distributed to different users for use, and the users can run various application programs in the corresponding client virtual machines. The client virtual machine can generate a data read-write request in the process of running the application program, and can call cloud storage service provided in the computing server to access the storage server cluster so as to perform read-write processing of data to be read-written.

The special processing chip includes, for example, the DPU illustrated in fig. 1, but is not limited thereto. The special processing chip is used for realizing the acceleration processing of the data read-write request triggered by the client virtual machine, so that other special processing chips capable of providing an acceleration computing function can be applicable. The general processing chip includes, for example, the CPU illustrated in fig. 1.

In practice, each computing server may include a plurality of general purpose processing chips (e.g., a plurality of CPUs, often referred to as a plurality of CPU cores) therein for constructing one or more guest virtual machines on the one hand and for providing the control plane functions described above on the other hand.

Compared with the traditional scheme of processing the data read-write request by using only a general processing chip, in the embodiment of the invention, the processing process of the data read-write request is divided into two processing processes of a data surface and a control surface, the control surface function mainly refers to the function of a network transmission layer and is used for performing network transmission control processing, and the data surface function mainly refers to data receiving and transmitting and related calculation processing based on the network transmission control processing result. The general processing chip provides control plane function, the special processing chip provides data plane function, the special processing chip (such as DPU) is limited by the programmable capability, is not suitable for processing complex states and logics (such as complex network transmission protocol processing work), but has better performance for simple logic processing than the general processing Chip (CPU), and the general processing chip is just opposite, and needs to interact with related devices such as a network card, a memory and the like when the general processing chip performs the data processing work, so that the processing time is longer. Based on this, in the embodiment of the present invention, only a small number of general processing chips need to be allocated to the computing server to provide the control plane function, and more general processing chips can be used to construct the client virtual machine for the user to use, thereby improving the utilization rate of the general processing chips.

Moreover, based on the architecture of separating the control plane from the data plane, logic in the cloud storage system, which is simple but needs to process a large amount of data, is separated to the data plane, and is implemented in a special processing chip (such as a DPU), so that the processing performance of I/O (input/output) can be improved, namely, the processing performance of data read-write requests can be improved. The routing, congestion control and retransmission mechanisms of the I/O are processed by the control plane, so that the design of the data plane can be simplified, and the complete network transmission layer function is realized.

In addition, by adopting the architecture with the control plane and the data plane separated, the control plane functions are irrelevant to the data, and better flexibility can be provided. For example, when a network protocol adopted between the computing server cluster and the storage server cluster needs to be updated, the general processing chip is mainly required to be adaptively updated, so that the cost is lower.

As shown in fig. 1, based on the architecture of separating the control plane and the data plane, a client virtual machine in a certain computing server may send a data read-write request to a dedicated processing chip that provides a data plane function, such as an NVMe request sent by using the NVMe specification as illustrated in fig. 1, based on an interface specification corresponding to a certain virtualization software. The control signal is transmitted between the special processing chip and the general processing Chip (CPU), so that the data content to be read and written is not carried, and the cost of a control surface is reduced. Based on the control surface processing result of the general processing chip, the special processing chip can transmit network messages with a certain storage server.

The following describes the operation of a special processing chip and a general processing chip in a certain computing server, respectively, with reference to the following embodiments.

Fig. 2 is an interaction flow chart of a cloud storage processing method provided by an embodiment of the present invention, as shown in fig. 2, the method includes the following steps:

201. the client virtual machine sends a data read-write request to the special processing chip, wherein the data read-write request comprises a logic disk storage address corresponding to data to be read-written.

202. The special processing chip sends a read-write request control signal containing a logic disk storage address to the general processing chip.

203. The general processing chip generates header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address.

204. The general processing chip transmits a message transmission control signal including header information to the special processing chip.

205. And the special processing chip generates a network request message according to the header information and the data read-write request.

206. The special processing chip sends the network request message to the target storage server so that the target storage server processes the data read-write request.

An application running in the guest virtual machine generates a data read-write request, and thus triggers the data read-write request. The data read-write request is a generic term for a data read request and a data write request.

In practical application, the communication between the special processing chip and the client virtual machine accords with a certain interface specification, such as Virtio, NVMe and other interface specifications, taking the NVMe specification as an example, the client virtual machine encapsulates the data read-write request triggered by the application program according to the interface specification, so as to obtain the NVMe request, and sends the NVMe request to the special processing chip.

The data read-write request includes a logical disk storage address corresponding to the data to be read-written, where the logical disk storage address refers to a logical block address (Logic Block Address, abbreviated as LBA). For a cloud storage service (generally referred to as a cloud disk service) provided by a cloud storage system, when a user uses the cloud storage service, an LBA address range corresponding to a certain capacity (for example, 1 TB) is allocated to the cloud storage service, and when a client virtual machine triggers a data read-write request, the client virtual machine needs to carry an LBA. Specifically, if the current data read-write request is a data read request, the LBA indicates a logical storage address corresponding to the data to be read in the storage server cluster, where the data to be read needs to be read from the address, and if the current data read-write request is a data write request, the LBA indicates that the data to be written needs to be written into the logical storage address in the storage server cluster.

In practice, the data read request also includes a memory address. For a data write request, the memory address indicates that the data to be written has been written to the memory address, and a dedicated processing chip is required to copy the data to be written from the memory address for final transmission to a storage server. For data read requests, the memory address indicates that data read from the storage server cluster needs to be written to the memory address by the dedicated processing chip in order for the guest virtual machine to retrieve the data therefrom.

After the special processing chip receives the NVMe request, the NVMe request is analyzed to extract a logic disk storage address from the NVMe request, a memory address can also be extracted, a read-write request control signal containing the logic disk storage address (or further comprising the memory address) is generated, and the read-write request control signal is sent to the general processing chip.

The general processing chip generates header information corresponding to the data read-write request based on a locally configured network transmission policy, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address, such as an IP address and a port number of the target storage server.

The network transmission policy may include a pre-configured routing table, where routing information corresponding to different logical disk storage addresses of different client virtual machines in a storage server cluster is recorded, where the routing information includes identification information of a storage server, and may also include a physical storage address corresponding to the storage server.

For example, assuming that the above data read/write request triggered by the guest virtual machine a includes a logical disk storage address of lba=x, it is determined according to the above routing table that the guest virtual machine a corresponds to the target storage server Z1, and the logical disk storage address of lba=x is mapped to the physical storage address Y, the message information generated by the general processing chip for the data read/write request includes identification information (IP address, etc.) of the target storage server Z1 and the physical storage address Y.

In addition, optionally, the network transmission policy may further include an I/O table, and the general processing chip may allocate, after receiving the read/write request control signal, read/write request identification information (referred to as I/O identification for convenience of description) corresponding to the read/write request control signal, where the I/O identification is used to uniquely identify a data read/write request. Because the control signal is transmitted between the special processing chip and the general processing chip, but not the original data, namely the data read-write request cannot be transmitted between the two chips, the I/O identifier is used as a representation for indicating that the control signal transmitted between the two subsequent chips is related to the same data read-write request.

After the general processing chip generates the I/O identifier, the I/O identifier may be stored in an I/O table. In practice, the information stored in the I/O table along with the I/O identifier may optionally further include a logical disk storage address, a memory address, and a reception time (indicating when the data read/write request is triggered) extracted by the general processing chip from the received read/write request control signal. Status information corresponding to the I/O identifier may also be recorded, where the status information is used to indicate what processing status the data read-write request is in, and includes, for example, a receiving status, a status of sending to the storage server, a status of receiving a network response message from the storage server, a status of not receiving the network response message after a timeout, and so on. When the I/O identifier is generated and stored in the I/O table, the initial state may be a receiving state, which indicates that a data read-write request triggered by the guest virtual machine is received.

In addition, the general processing chip may also make determinations such as network transmission parameters, such as data encryption parameters, checksum parameters, and the like, based on the network transmission control policy.

In summary, header information generated by the general processing chip for the data read-write request based on the network transmission control policy includes, but is not limited to, identification information of the target storage server, I/O identification and other network transmission parameters, and even may further include the extracted memory address. The general processing chip organizes the generated header information into a message sending control signal and sends the message sending control signal to the special processing chip.

It should be noted that, in the process of sending the above-mentioned message sending control signal to the special processing chip, the general processing chip may determine a congestion state of a network connection between the computing server and the target storage server, and determine, according to the congestion state, a sending time of the message sending control signal, so as to send the message sending control signal to the special processing chip according to the sending time. In short, if the network connection is currently in a non-congestion state, the message sending control signal can be sent to the special processing chip immediately, otherwise, the judgment of whether the network connection is in a congestion state can be performed after waiting for a set time period. That is, the network congestion control is also included in the network transmission control policy of the general processing chip.

And after receiving the message sending control signal, the special processing chip encapsulates the network request message according to the header information carried in the message sending control signal.

For example, if the data read-write request is a data write request, the data write request includes a memory address corresponding to the data to be written, so that optionally, the read-write request control signal and the message sending control signal may also include the memory address. At this time, the special processing chip may read the data to be written from the memory address based on the DMA, and generate a network request message according to the header information and the data to be written, where a header portion of the network request message includes information such as the identification information, the I/O identification, and the like of the target storage server. When the header information contains data encryption parameters, the special processing chip also encrypts data to be written based on the data encryption parameters, and the encryption result is filled in the load part of the network request message.

And the special processing chip sends the packaged network request message to the target storage server so as to enable the target storage server to perform corresponding data read-write processing.

In summary, based on the architecture of separating the control plane from the data plane, the processing logic is simple, but the logic needed to process a large amount of data is separated to the data plane, the processing logic is realized in a special processing chip, the complex network transmission control function is separated to the control plane, the processing of the control plane function can be carried out in a general processing chip, only a small amount of general processing chips can be allocated to a calculation server to be responsible for the processing of the control plane function, a small amount of special processing chips are deployed to process the data plane function, the overall cost is lower, and the special processing chips can realize the acceleration processing of the data plane function (for example, the data processing chip can only need to read or write the memory address once because of integrating the network card function), the processing time consumption of the data read-write request can be reduced, and the throughput performance is improved.

The above description is given of the process that the special processing chip and the general processing chip send the data read-write request to the storage server cluster, and the following description is given to the processing process of the feedback network response message of the target storage server by combining the embodiments shown in fig. 3 and fig. 4.

Fig. 3 is an interaction flow chart of a cloud storage processing method provided by an embodiment of the present invention, as shown in fig. 3, the method includes the following steps:

301. The target storage server sends a network response message to the special processing chip, wherein the network response message comprises read-write request identification information and first message type information corresponding to the network response message.

302. The special processing chip sends a first message response control signal containing read-write request identification information and first message type information to the general processing chip.

303. The general processing chip generates a read-write completion control signal according to the first message type information, wherein the read-write completion control signal comprises read-write request identification information.

304. The general processing chip sends the read-write completion control signal to the special processing chip.

305. The special processing chip generates a data read-write response containing read-write request identification information.

306. The special processing chip sends the data read-write response to the client virtual machine.

After receiving the network request message, the target storage server executes corresponding data read-write processing, and then generates a network response message to be fed back to the special processing chip.

It can be understood that if the client virtual machine triggers a data write request, the target storage server completes the writing process of the data to be written, and if the client virtual machine triggers a data read request, the target storage server completes the reading process of the read data, and at this time, the network response message carries the read data.

The header portion of the network response message generated by the target storage server may carry the I/O identifier and the first message type information, and may further include identification information, such as identification information of the target storage server. The first message type information is used for representing that a message of the type of a network response message is sent, and the type information indicates that the target storage server has successfully completed corresponding data read-write processing.

After receiving the network response message, the special processing chip analyzes the network response message to extract the I/O identification and the first message type information from the network response message, generates a first message response control signal containing the I/O identification and the first message type information, and sends the first message response control signal to the general processing chip to inform the general processing chip that the general processing chip has received the network response message fed back by the target storage server.

The general processing chip determines that the special processing chip has received the network response message fed back by the target storage server based on the first message type information. So that a read-write complete control signal can be generated and sent to the dedicated processing chip. In practice, the general processing chip may update congestion status information of the network connection between the computing server and the target storage server based on the received first packet response control signal.

And after receiving the read-write completion control signal, the special processing chip performs data read-write response processing, generates data read-write response containing the I/O identifier, and sends the data read-write response to the client virtual machine. If the client virtual machine triggers a data read request, the data read-write response processing includes responding to the read-write completion control signal and storing read data into a memory address, wherein the memory address is a memory address contained in the data read-write request.

Specifically, assuming that an interface specification used between the dedicated processing chip and the client virtual machine is NVMe, the data read-write response generated by the dedicated processing chip is an NVMe response, and may further include the memory address for the data read request, so that the client virtual machine will read the data in the memory address.

The embodiment shown in fig. 3 above describes the situation that the target storage server finally sends the network response message to the special processing chip after completing the data read-write process, in fact, before sending the network response message, the target storage server may also send another type of network message, namely, a receipt acknowledgement message, for indicating that the network request message sent by the special processing chip has been received. Based on the receipt confirmation message, the general processing chip can perform network retransmission control, referring to the embodiment shown in fig. 4.

Fig. 4 is an interaction flow chart of a cloud storage processing method provided by an embodiment of the present invention, as shown in fig. 4, the method includes the following steps:

401. The target storage server sends a receiving confirmation message to the special processing chip, wherein the receiving confirmation message comprises read-write request identification information, second message type information corresponding to the receiving confirmation message and message sequence number information.

402. The special processing chip sends a second message response control signal containing the read-write request identification information, the second message type information and the message serial number information to the general processing chip.

403. And regenerating message sending control signals corresponding to other read-write request control signals by the universal processing chip according to the second message response control signals, wherein the other read-write request control signals correspond to other data read-write requests before the data read-write request.

404. The general processing chip sends the regenerated message sending control signal to the special processing chip.

405. And the special processing chip sends a control signal according to the regenerated message, and regenerates the corresponding network request message.

406. And the special processing chip sends the regenerated network request message to a corresponding storage server.

For convenience of description, in this embodiment, it is assumed that the same guest virtual machine triggers the data read-write request 1, the data read-write request 2 and the data read-write request 3 before and after, and the transmission flow of each data read-write request may refer to the processing procedure of the embodiment shown in fig. 2, and it is assumed that the general processing chip sequentially sends corresponding packet transmission control signals to the dedicated processing chip for the three data read-write requests, where the target storage servers corresponding to the three data read-write requests are assumed to be the same.

And the special processing chip generates three corresponding network request messages according to the three message transmission control signals received in sequence and sequentially transmits the three corresponding network request messages to the target storage server. The target storage server generates a corresponding second message response control signal and sends the second message response control signal to the special processing chip after receiving a network request message. The second message response control signal contains a message sequence number corresponding to the corresponding network request message. The special processing chip compares the size sequence of the message sequence numbers in the sequentially received second message response control signals to determine whether certain network request messages need to be retransmitted or not.

The method comprises the steps that a data read-write request 1 corresponds to a network request message 1, the sending time is T1, a data read-write request 2 corresponds to a network request message 2, the sending time is T2, a data read-write request 3 corresponds to a network request message 3, and the sending time is T3. For ease of description, it is assumed that the time consumed by the dedicated processing chip to send the above network request message to the target storage server is ignored.

It is assumed that the target storage server currently receives the network request message 3, so as to generate a receipt acknowledgement message 3 corresponding to the network request message 3, where the receipt acknowledgement message 3 includes an I/O identifier carried in the network request message 3, second message type information indicating that the message is a receipt acknowledgement message, and message sequence number information (in this example, sequence number 3) corresponding to the network request message 3.

After the general processing chip receives the second message response control signals containing the information, the target storage server is determined to receive the network request message 3, but the general processing chip discovers that the second message response control signals corresponding to the network request message 1 and the network request message 2 are not received based on the received second message response control signals, and the general processing chip indicates that the network request message 1 and the network request message 2 are not received by the target storage server, namely the network request message 1 and the network request message 2 are lost. And at the moment, the general processing chip triggers retransmission processing of the two network request messages, namely, based on the read-write request control signals corresponding to the two corresponding data read-write requests received before, regenerating the corresponding message transmission control signals and sending the message transmission control signals to the special processing chip, so that the special processing chip regenerates the two network request messages according to the received message transmission control signals and sends the two network request messages to the target storage server.

In summary, in the embodiment of the invention, the cloud storage system is disassembled into a framework with a control surface and a data surface separated, the data surface is realized in a special processing chip, and the control surface is realized in a general processing chip. The special processing chip has limited programming capability and is not suitable for processing complex states and logic, but the performance of the special processing chip is superior to that of the general processing chip for simple processing logic, and the processing performance of I/O can be improved. The provided control plane of the I/O and transmission layer which is irrelevant to the data can realize complete network transmission layer functions with congestion control and reliable transmission capability, and the routing, congestion control and retransmission of the I/O are processed by the control plane, so that the flexibility and the reliability are high, and the design of the data plane is simplified. Only control signals are transmitted between the data plane and the control plane, no data is carried, and the cost of the control plane is reduced. The construction of the header information is accomplished by the control plane and can be switched to a different network transport layer protocol at low cost.

The above description is made for the cooperation process of the general processing chip and the special processing chip, and the following description is made for the composition structure of the data plane function on the special processing chip and the composition structure of the control plane function on the general processing chip, respectively.

Fig. 5 is a schematic diagram of a composition structure of a data plane according to an embodiment of the present invention. As shown in fig. 5, the client virtual machine triggers a data read-write request, i.e. an I/O request, specifically, it is assumed that an I/O request triggered by an NVMe specification (referred to as an NVMe request) will first enter an NVMe request queue, and the NVMe request processing module will consume the NVMe request queue, i.e. sequentially read out one NVMe request from the NVMe request, extract I/O information (such as LBA and memory address) therein, generate an I/O request control signal (i.e. the above read-write request control signal), and send the I/O request control signal to the control signal sending module for sending to the control plane by the control signal sending module.

After receiving the message sending control signal sent by the control surface, the control signal receiving module sends the message sending control signal to the network message assembling module. The network message assembly module constructs a network message header according to the message sending control signal, acquires data to be written (assuming that the NVMe request is a data writing request) through DMA, assembles and generates a network request message, and sends the network request message to the message sending module so as to be sent to the target storage server by the message sending module. After the network response message of the target storage server reaches the message receiving module of the data surface, the network response message is sent to the network message analyzing module, and after the network message analyzing module determines that the received network message accords with the set characteristic (for example, the received first message type information indicating that the message type is the network response message), a message response control signal is generated, and the message response control signal is sent to the control signal sending module so as to be sent to the control surface by the control signal sending module. After receiving the I/O completion control signal (i.e., the read-write completion control signal above) sent by the control plane, the control signal receiving module sends the I/O completion control signal to the NVMe response processing module. The NVMe response processing module constructs an NVMe request completion command (i.e., the data read-write response above) and sends the command to the NVMe request completion queue.

Fig. 6 is a schematic diagram of a composition structure of a control surface according to an embodiment of the present invention. As shown in fig. 6, the control plane receives various control signals from the data plane from the control signal receiving module.

For I/O request control signals, the I/O processing module is entered. The I/O processing module checks the I/O request, assigns a unique identifier of the I/O request, and writes the unique identifier into an I/O table. The I/O request then enters the routing table to find the target storage server to which the I/O request corresponds. And finally, the I/O request enters a transmission layer control module to perform congestion control, and a message sending control signal is generated to be sent to a data surface through a control signal sending module. The I/O request control signal comprises an LBA and a memory address, and the checking of the I/O request comprises determining whether the data quantity to be read and written is consistent according to the two addresses, and determining whether the data quantity to be read and written exceeds the upper limit of the usable capacity of the client virtual machine. For example, if 500MB of data is to be written, the address ranges described by the two addresses should both be consistent with 500MB of data, and the storage space available to the guest virtual machine should be greater than 500 MB.

And responding to the control signal by the message, and entering a message processing module. After the message processing module analyzes the message response control signal, the message response control signal is sent to the transmission layer control module. The transmission layer control module judges the type of the network message and judges the retransmission. And finally, the message response control signal enters an I/O table, finds corresponding I/O information according to the I/O identifier, and generates an I/O completion signal to be sent to a data surface through a control signal sending module.

The following illustrates the processing procedure of the transport layer control module in the control plane in the network message sending and receiving procedure in connection with fig. 7 and 8, respectively.

Fig. 7 is a schematic diagram of a processing procedure of a control plane in a network message sending process according to an embodiment of the present invention. As shown in fig. 7, the I/O request control signal, the retransmission signal, and the window update signal all trigger the transmission procedure of the transport layer control module.

Specifically, for the I/O request control signal received from the data plane, the I/O request control signal is added to the transmission queue first, then the congestion state of the network connection is checked, if the network connection is congested, the next judgment is directly finished (after waiting for a set period of time), otherwise, the transmission layer control module consumes the transmission queue, i.e. takes out an I/O request control signal therefrom, generates header information, generates a message transmission control signal and transmits the message transmission control signal to the data plane, and updates the connection table.

The transport layer control module may maintain a connection table, where congestion status information of network connections between different computing servers and storage servers may be recorded, for example, how many message sending control signals (assuming that N1 messages are currently sent) have been sent on a network connection, and how many message response control signals (assuming that N2 messages are currently sent) indicating that acknowledgement messages are received. Based on this, the congestion determination may be implemented by determining whether the difference between N1-N2 is greater than a set threshold, and if so, indicating that the network connection between the corresponding computing server and the storage server is in a congested state, otherwise in a non-congested state.

The updating of the connection table may be simply understood as adding one to the value of N1 corresponding to the network connection between the corresponding computing server and the storage server after the sending of the message transmission control signal.

Fig. 8 is a schematic diagram of a processing procedure of a control plane in a network packet receiving process according to an embodiment of the present invention. As shown in fig. 8, after the transport layer control module receives a certain message response control signal fed back by the data plane, it first determines whether it is a reception acknowledgement message. Specifically, as described above, the message type information is determined according to the message type information carried in the message response control signal. If not, indicating that a network response message is received, generating a window update signal. If yes, checking whether the message is out of order. The window update signal is to generate a consume signal to consume an I/O request control signal waiting in the transmit queue.

If not, the connection table is updated directly. The update of the connection table is, for example, to increase the value of N2 corresponding to the network connection between the corresponding computing server and the storage server by one.

If the messages are out of order, the I/O request control signals corresponding to the network request messages which are not confirmed to be received are added into the transmission queue shown in fig. 7 again, retransmission signals are generated to retransmit the I/O request control signals, finally, a connection table is updated, for example, the value of the N1 is updated according to retransmission conditions, and the size of the set threshold value compared with the difference value of the N1-N2 can be adjusted.

Communication devices of one or more embodiments of the present invention will be described in detail below. Those skilled in the art will appreciate that these means may be configured by the steps taught by the present solution using commercially available hardware components.

Fig. 9 is a schematic structural diagram of a cloud storage processing device according to an embodiment of the present invention, where the device is applied to the above-mentioned special processing chip, and as shown in fig. 9, the device includes a receiving module 11, a sending module 12, and a generating module 13.

The receiving module 11 is configured to receive a data read-write request sent by a client virtual machine, where the data read-write request includes a logical disk storage address corresponding to data to be read-written.

And a sending module 12, configured to send a read-write request control signal containing the logical disk storage address to a general processing chip configured to provide a control plane function in the computing server, so that the general processing chip generates header information, where the header information includes identification information of a target storage server corresponding to the logical disk storage address.

The receiving module 11 is further configured to receive a message sending control signal sent by the general processing chip, where the message sending control signal includes the header information.

And the generating module 13 is used for generating a network request message according to the header information and the data read-write request.

The sending module 12 is further configured to send the network request packet to the target storage server, so that the target storage server processes the data read-write request.

The apparatus shown in fig. 9 may perform the steps performed by the processing chip in the foregoing embodiment, and the detailed performing process and technical effects are referred to the description in the foregoing embodiment, which is not repeated herein.

Fig. 10 is a schematic structural diagram of a cloud storage processing device according to an embodiment of the present invention, where the device is applied to the above general processing chip, and as shown in fig. 10, the device includes a receiving module 21, a generating module 22, and a sending module 23.

The receiving module 21 is configured to receive a read-write request control signal sent by a dedicated processing chip configured to provide a data plane function in the computing server, where the read-write request control signal is sent after the dedicated processing chip receives a data read-write request sent by a client virtual machine, the data read-write request includes a logical disk storage address corresponding to data to be read-written, and the read-write request control signal includes the logical disk storage address.

And the generating module 22 is configured to generate header information, where the header information includes identification information of the target storage server corresponding to the logical disk storage address.

The sending module 23 is configured to send a message sending control signal containing the header information to the dedicated processing chip, so that the dedicated processing chip generates a network request message according to the header information and the data read-write request, and send the network request message to the target storage server, so that the target storage server processes the data read-write request.

The apparatus shown in fig. 10 may perform the steps performed by the general processing chip in the foregoing embodiments, and detailed execution and technical effects are referred to in the foregoing embodiments and are not described herein.

The embodiment of the invention also provides a computing server, which can comprise a special processing chip 31, a general processing chip 32, a memory 33 and a communication interface 34 as shown in fig. 11. In which executable code is stored on the memory 33, which when executed by the dedicated processing chip 31, the general-purpose processing chip 32, implements the cloud storage processing method as in the foregoing embodiment.

In addition, an embodiment of the present invention provides a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a special-purpose processing chip and a general-purpose processing chip in a computing server, can implement the cloud storage processing method as provided in the foregoing embodiment.

The apparatus embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by adding necessary general purpose hardware platforms, or may be implemented by a combination of hardware and software. Based on such understanding, the foregoing aspects, in essence and portions contributing to the art, may be embodied in the form of a computer program product, which may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

It should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention, and not for limiting the same, and although the present invention has been described in detail with reference to the above-mentioned embodiments, it should be understood by those skilled in the art that the technical solution described in the above-mentioned embodiments may be modified or some technical features may be equivalently replaced, and these modifications or substitutions do not make the essence of the corresponding technical solution deviate from the spirit and scope of the technical solution of the embodiments of the present invention.

Claims (14)

1. A cloud storage processing method, applied to a dedicated processing chip configured to provide a data plane function in a computing server, the method comprising:

receiving a data read-write request sent by a client virtual machine, wherein the data read-write request comprises a logic disk storage address corresponding to data to be read-written;

Transmitting a read-write request control signal containing the logical disk storage address to a general processing chip configured to provide a control plane function in the computing server, so that the general processing chip generates header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address;

receiving a message sending control signal sent by the general processing chip, wherein the message sending control signal comprises the header information;

Generating a network request message according to the header information and the data read-write request, and sending the network request message to the target storage server so that the target storage server processes the data read-write request.

2. The method of claim 1, wherein the message sending control signal further includes read-write request identification information allocated by the general processing chip and corresponding to the read-write request control signal, and the network request message includes the read-write request identification information.

3. The method of claim 1, wherein the data read-write request is a data write request, and the data write request includes a memory address corresponding to data to be written;

the generating a network request message according to the header information and the data read-write request comprises the following steps:

Reading the data to be written from the memory address;

And generating a network request message according to the header information and the data to be written.

4. The method according to claim 2, wherein the method further comprises:

Receiving a network response message sent by the target storage server, wherein the network response message comprises the read-write request identification information and first message type information corresponding to the network response message;

transmitting a first message response control signal containing the read-write request identification information and the first message type information to the general processing chip;

receiving a read-write completion control signal generated by the general processing chip according to the first message type information, wherein the read-write completion control signal comprises the read-write request identification information;

generating a data read-write response containing the read-write request identification information, and sending the data read-write response to the client virtual machine.

5. The method of claim 4, wherein the data read-write request is a data read request, the data read request includes a memory address corresponding to read data;

The method further comprises the steps of:

and responding to the read-write completion control signal, and storing the read data into the memory address.

6. The method according to claim 2, wherein the method further comprises:

Receiving a receiving confirmation message sent by the target storage server, wherein the receiving confirmation message comprises the read-write request identification information, second message type information corresponding to the receiving confirmation message and message sequence number information;

Transmitting a second message response control signal containing the read-write request identification information, the second message type information and the message sequence number information to the general processing chip;

Receiving message sending control signals corresponding to other read-write request control signals regenerated by the general processing chip according to the second message response control signals, wherein the other read-write request control signals correspond to other data read-write requests before the data read-write request;

and according to the regenerated message sending control signal, regenerating a corresponding network request message and sending the network request message to a corresponding storage server.

7. A cloud storage processing method, applied to a general processing chip configured to provide a control plane function in a computing server, the method comprising:

Receiving a read-write request control signal sent by a special processing chip configured to provide a data surface function in the computing server, wherein the read-write request control signal is sent after the special processing chip receives a data read-write request sent by a client virtual machine, the data read-write request comprises a logic disk storage address corresponding to data to be read-written, and the read-write request control signal comprises the logic disk storage address;

Generating header information, wherein the header information comprises identification information of a target storage server corresponding to the logical disk storage address;

and sending a message sending control signal containing the header information to the special processing chip so that the special processing chip generates a network request message according to the header information and the data read-write request, and sending the network request message to the target storage server so that the target storage server processes the data read-write request.

8. The method of claim 7, wherein the method further comprises:

and distributing the read-write request identification information corresponding to the read-write request control signal, wherein the message sending control signal and the network request message contain the read-write request identification information.

9. The method of claim 7, wherein transmitting a message transmission control signal containing the header information to the dedicated processing chip comprises:

Determining a congestion state of a network connection between the computing server and the target storage server;

Determining the sending time of the message sending control signal according to the congestion state;

and transmitting the message transmission control signal to the special processing chip according to the transmission time.

10. The method of claim 8, wherein the method further comprises:

Receiving a first message response control signal sent by the special processing chip, wherein the first message response control signal comprises the read-write request identification information and first message type information corresponding to a network response message, the first message response control signal is sent after the special processing chip receives the network response message which is sent by the target storage server and corresponds to the network request message, and the network response message comprises the read-write request identification information and the first message type information;

generating a read-write completion control signal according to the first message type information, wherein the read-write completion control signal comprises the read-write request identification information;

And sending the read-write completion control signal to the special processing chip so that the special processing chip generates a data read-write response containing the read-write request identification information, and sending the data read-write response to the client virtual machine.

11. The method of claim 8, wherein the method further comprises:

Receiving a second message response control signal sent by the special processing chip, wherein the second message response control signal comprises the read-write request identification information, second message type information and message sequence number information corresponding to a receiving confirmation message, the second message response control signal is sent after the special processing chip receives the receiving confirmation message which is sent by the target storage server and corresponds to the network request message, and the receiving confirmation message comprises the read-write request identification information, the second message type information and the message sequence number information;

If the condition that the network request message is out of order is determined according to the second message type information and the message sequence number information, regenerating a message sending control signal corresponding to other read-write request control signals, and sending the regenerated message sending control signal to the special processing chip so as to enable the special processing chip to regenerate the corresponding network request message, wherein the other read-write request control signals correspond to other data read-write requests before the data read-write request.

12. A computing server, comprising a memory, a general purpose processing chip, a special purpose processing chip, and a communication interface, wherein the memory stores executable code thereon, and when the executable code is executed by the general purpose processing chip and the special purpose processing chip, the cloud storage processing method according to any one of claims 1 to 6 is executed, or the cloud storage processing method according to any one of claims 7 to 11 is executed.

13. A non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a general purpose processing chip and a special purpose processing chip in a computing server, performs the cloud storage processing method of any of claims 1 to 6 or performs the cloud storage processing method of any of claims 7 to 11.

14. A cloud storage system, comprising:

A computing server cluster and a storage server cluster, wherein any computing server in the computing server cluster comprises a client virtual machine, a special processing chip configured to provide a data plane function and a general processing chip configured to provide a control plane function;

The dedicated processing chip for executing the cloud storage processing method according to any one of claims 1 to 6;

The general processing chip is configured to perform the cloud storage processing method according to any one of claims 7 to 11.