High-speed Ethernet and computing power fused data processing unit

Abstract

The invention relates to a high-speed Ethernet and computing power fused data processing unit. The system comprises an FPGA, a DDR interface, a Flash interface, four internal data exchange paths and two external data exchange paths, the DDR interface, the Flash interface, the four internal data exchange paths and the two external data exchange paths are respectively connected with the FPGA, and the FPGA realizes a logic processing function; the DDR interface is connected with the DDR memory and serves as a CPU temporary storage space; the Flash interface is responsible for loading the logic unit; the four internal data exchange paths are respectively connected with four hosts HOST0, HOST1, HOST2 and HOST3, and communication among different hosts is realized through an FPGA (Field Programmable Gate Array); the two external data exchange paths are connected with external equipment, and communication between different hosts and the external equipment is achieved through the FPGA. According to the invention, a plurality of hosts can be supported, the hosts can share the computing power of the data processing unit, internal exchange and external exchange of data can be realized, the data throughput is increased, the reliability and safety of the system are improved, and the airborne communication scene is better met.

Description

Technical Field

[0001] The present invention belongs to the field of communication equipment, and in particular relates to a data processing unit integrating high-speed Ethernet and computing power. Background Art

[0002] The current airborne transmission unit is mainly a computing model with the CPU as the core. However, there are many disadvantages in using the CPU as the computing core. The CPU is originally used for complex control. Using the CPU to perform complex calculations will waste CPU resources. Even if the CPU is used as the core for calculations, the CPU interface is limited and the computing power is difficult to promote. And with the advent of the big data era, the CPU-based computing model has poor data exchange capabilities, insufficient computing and storage capabilities, and low utilization when facing massive big data and highly dynamic intelligent service calculations due to the doubling of the number of services and computing power. There are also problems such as inflexible and unbalanced resource allocation, difficulty in configuring resources on demand, and insufficient data throughput and real-time performance, which cannot meet the requirements of airborne network data processing.

[0003] At present, some network processing units and transmission units only serve one host. However, a single-host environment brings many limitations. The computing resources of a single host are limited and cannot meet the needs of processing large-scale data, resulting in performance bottlenecks. The failure of a single host will cause the entire system to be unavailable, increasing the risk of the onboard system. In a single-host environment, various services share the resources of the host, which will lead to resource competition. Therefore, a single host has many limitations. Especially when facing large-scale data processing, high availability requirements and continuous expansion needs, a multi-host environment is often more suitable for meeting these challenges and improving the performance, reliability and flexibility of the system. Summary of the invention

[0004] In order to solve the technical problem that the current network processing unit can only provide limited data processing capabilities and cannot meet complex data processing requirements, the present invention provides a data processing unit that integrates high-speed Ethernet and computing power. The data processing unit can support multiple hosts, and multiple hosts can share the computing power of the data processing unit, and can realize internal and external data exchange, increase data throughput, improve system reliability and security, and better meet airborne communication scenarios.

[0005] The technical solution of the present invention is: the present invention is a data processing unit integrating high-speed Ethernet and computing power, and its special support is: the data processing unit integrating high-speed Ethernet and computing power includes FPGA, DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths, the DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths are respectively connected to the FPGA, and the FPGA realizes the logic processing function; the DDR interface is connected to the DDR memory as a temporary storage space of the CPU; the Flash interface is responsible for loading the logic unit; the four internal data exchange paths are respectively connected to four hosts HOST0, HOST1, HOST2, and HOST3, and communication between different hosts is realized through the FPGA; the two external data exchange paths are connected to external devices, and communication between different hosts and external devices is realized through the FPGA.

[0006] Furthermore, each host supports 4 functions, with a total of 16 functions. Each function is a network port, each network port has its own IP address, and each IP address can run a variety of different network applications. Different hosts and different functions are isolated from each other and do not interfere with each other.

[0007] Furthermore, AF is deployed in HOST0. AF is used to manage the hardware resources of the entire processing unit. Only AF can access the hardware resources. PF in HOST0 applies for hardware resources from PF in HOST1, HOST2, and HOST3. It needs to communicate with AF in HOST0, and AF decides whether to configure it.

[0008] Furthermore, the business traffic of each host can be allocated on demand according to actual business needs to actively control the total traffic volume and traffic priority of each host. Each function within the host can also be configured with the total traffic volume and traffic priority of each function according to business needs, and the number of queues that each function can use can also be configured on demand.

[0009] Furthermore, data is forwarded through the stream forwarding engine inside the data processing unit; the internal logic of the data processing unit has a variety of stream forwarding modes, and different stream forwarding modes can be selected as needed; these forwarding modes are associated with the identification information and MAC address information of each function. When different functions communicate with each other or with external devices, the data processing unit will determine the target based on this information, thereby accurately forwarding the data to the target address.

[0010] A data processing method of the data processing unit integrating high-speed Ethernet and computing power is special in that the method comprises the following steps:

[0011] 1) Internal forwarding: When traffic comes out of Function 1 of HOST1, it is encapsulated by the protocol stack and then reaches the data processing unit through the driver. The data processing unit determines that the destination of the current data packet is Function 1 of HOST2 based on the internal flow forwarding engine logic. The data packet is then sent to the driver of Function 1 of HOST2 and then decapsulated by the protocol stack and sent to the user's upper-layer application. This process is repeated in this way to achieve internal forwarding.

[0012] 2) External forwarding: The destination is an external device. Function 1 of HOST1 sends data to the external device. When the traffic comes out of Function 1 of HOS1, it is encapsulated by the protocol stack and then driven to the data processing unit. The data processing unit determines that the destination of the current data packet is the external device of the data processing unit based on the internal stream forwarding engine logic, so the data packet will be sent directly to the Ethernet port and then to the external device.

[0013] The present invention proposes a data processing unit that integrates high-speed Ethernet and computing power. The data processing unit can support multiple hosts, and different services run on different hosts. The services of different hosts can exchange internal data in the unit, and the services of all hosts can also exchange external data with external devices through the Ethernet port of the unit. In this way, while releasing the CPU computing power, the unit increases the data throughput, can process more requests at the same time, avoids resource competition and influence, improves the stability and security of the system, and solves the problem of mismatch between existing network card computing resources, storage resources and network bandwidth. In addition, the architecture of the unit can freely select functional nodes, realize on-demand configuration and on-demand allocation of resources, thereby greatly improving the airborne intelligent application capabilities and better meeting airborne communication and computing needs.

[0014] The present invention provides a new solution to the problems of poor data exchange capability, insufficient computing and storage capacity, and low utilization rate of existing airborne transmission units when facing massive big data and highly dynamic intelligent service calculations due to the doubling of the number of services and computing power. The transmission unit can be directly connected to the network, and when connected to an external computing core, the external computing core can control the computing platform. At the same time, the invention can support the business needs of multiple hosts at the same time, forming a processing unit with aggregated computing power, increasing data throughput while freely selecting functional nodes to realize on-demand configuration and on-demand allocation of resources. It can meet more airborne business needs and make the system more secure and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a structural block diagram of the present invention;

[0016] Figure 2 This is a configuration diagram of the multi-host computing power aggregation data processing unit of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] See also Figure 1 The structure of the specific embodiment of the present invention includes FPGA, DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths. The DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths are connected to the FPGA respectively, and the DDR interface is connected to the DDR memory; the Flash interface is responsible for loading the logic unit; the four internal data exchange paths are connected to the four hosts HOST0, HOST1, HOST2, and HOST3 respectively, and the two external data exchange paths are connected to the external device. Among them:

[0019] As the main part of the unit, FPGA realizes the logic processing function;

[0020] The DDR interface is connected to the DDR memory as temporary storage space for the CPU;

[0021] Four internal data exchange paths enable communication between different hosts through FPGA;

[0022] 2-way external data exchange path realizes the communication between the host and external devices through FPGA;

[0023] The Flash interface is responsible for loading the logic unit.

[0024] The data processing unit can support multiple hosts at the same time. Different hosts can exchange internal data through the data processing unit. The data processing unit also has a high-speed external interface through which the host can exchange external data with external devices. The data processing unit serves multiple hosts and constitutes a multi-host computing power aggregation data processing unit configuration diagram, as shown in Figure 2 As shown:

[0025] Multiple hosts share the same data processing unit, forming a multi-host computing power aggregation. The FPGA inside the data processing unit is the core unit of this configuration, executing logical functions, processing complex operations, and then transmitting data through four internal data exchange channels and two external data exchange channels.

[0026] This configuration supports a total of 4 hosts (HOST): HOST0, HOST1, HOST2, HOST3, each HOST has 4 different functions (Function), these functions are independent of each other, and share the same HOST. Among them, AF is deployed in HOST0. AF is a special function used to manage the hardware resources of the entire processing unit. Only AF can access the hardware resources. If other PFs in HOST0 want to apply for hardware resources with PFs of other HOSTs, they need to communicate with AF of HOST0, and AF decides whether to configure it.

[0027] Each Function of each HOST can run multiple different services. Different HOSTs or different Functions of the same HOST can communicate through internal switching channels. Similarly, each Function can also communicate with external devices through external switching channels.

[0028] The architecture of this configuration can freely select functional nodes to realize on-demand resource configuration and allocation. In the figure, the business traffic of each HOST can be allocated on demand according to actual business needs to actively control the total traffic volume and traffic priority of each HOST. Each Function in the HOST can also configure the total traffic volume and traffic priority of each Function according to business needs, and the number of queues that each Function can use can also be configured on demand.

[0029] Data is forwarded mainly through the flow forwarding engine inside the processing unit. The internal logic of the processing unit has multiple flow forwarding modes, and different flow forwarding modes can be selected as needed. These forwarding modes are associated with the identification information, MAC address information, etc. of each Function. When different Functions communicate with each other or Functions communicate with external devices, the processing unit will determine the target based on this information, thereby accurately forwarding the data to the target address.

[0030] The data processing unit supports 4 HOSTs, each HOST supports 4 Functions, that is, there are 16 Functions in total. Each Function is a network port, each network port has its own IP address, and each IP address can run a variety of different network applications. Different HOSTs and different Functions are isolated from each other and do not interfere with each other.

[0031] The present invention also provides a data processing method integrating high-speed Ethernet and computing power network, and the working steps of the specific embodiment are as follows:

[0032] 1) For internal forwarding. This process is described by the transmission of Function 1 of HOST1 and the reception of Function 1 of HOST2. When the traffic comes out of Function 1 of HOST1, it is encapsulated by the protocol stack and then reaches the data processing unit through the driver. The data processing unit determines that the destination of the current data packet is Function 1 of HOST2 based on the internal flow forwarding engine logic, so the data packet is sent to the driver of Function 1 of HOST2, and then decapsulated by the protocol stack and sent to the user's upper-layer application.

[0033] 2) For external forwarding. The destination is an external device. This process is described by Function 1 of HOST1 sending data to an external device. When the traffic comes out of Function 1 of HOST1, it is encapsulated by the protocol stack and then driven to the data processing unit. The data processing unit determines that the destination of the current data packet is the external device of the data processing unit based on the internal flow forwarding engine logic, so the data packet will be directly sent to the Ethernet port and then to the external device.

[0034] The above are only specific embodiments disclosed in the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

[0035] The content of the present invention and the technical content not specifically described in the above embodiments are the same as the prior art.

[0036] The present invention is not limited to the above embodiments, and all of the contents of the present invention can be implemented and have the above good effects.

Claims

1. A data processing unit integrating high-speed Ethernet and computing power, characterized in that: The data processing unit integrating high-speed Ethernet and computing power includes FPGA, DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths. The DDR interface, Flash interface, four internal data exchange paths and two external data exchange paths are respectively connected to the FPGA, and the FPGA implements the logic processing function; the DDR interface is connected to the DDR memory as a temporary storage space of the CPU; the Flash interface is responsible for loading the logic unit; the four internal data exchange paths are respectively connected to four hosts HOST0, HOST1, HOST2, and HOST3, and communication between different hosts is realized through the FPGA; the two external data exchange paths are connected to external devices, and communication between different hosts and external devices is realized through the FPGA.

2. The data processing unit integrating high-speed Ethernet and computing power according to claim 1, characterized in that: Each host supports 4 functions, with a total of 16 functions. Each function is a network port. Each network port has its own IP address. Each IP address can run a variety of different network applications. Different hosts and different functions are isolated from each other and do not interfere with each other.

3. The data processing unit integrating high-speed Ethernet and computing power according to claim 2, characterized in that: AF is deployed in HOST0. AF is used to manage the hardware resources of the entire processing unit. Only AF can access the hardware resources. PF in HOST0 applies for hardware resources from PF in HOST1, HOST2, and HOST3. It needs to communicate with AF in HOST0, and AF decides whether to configure it.

4. The data processing unit integrating high-speed Ethernet and computing power according to claim 3 is characterized in that: The business traffic of each host can be allocated on demand according to actual business needs to actively control the total traffic volume and traffic priority of each host. The total traffic volume and traffic priority of each function in the host can also be configured according to business needs. The number of queues that each function can use can also be configured on demand.

5. The data processing unit integrating high-speed Ethernet and computing power according to claim 3, characterized in that: Data is forwarded through the flow forwarding engine inside the data processing unit. The internal logic of the data processing unit has multiple flow forwarding modes, and different flow forwarding modes can be selected as needed. These forwarding modes are associated with the identification information and MAC address information of each function. When different functions communicate with each other or with external devices, the data processing unit will determine the target based on this information, thereby accurately forwarding the data to the target address.

6. A data processing method of a data processing unit integrating high-speed Ethernet and computing power according to claim 1, characterized in that: The method comprises the following steps: 1) Internal forwarding: When traffic comes out of Function 1 of HOST1, it is encapsulated by the protocol stack and then reaches the data processing unit through the driver. The data processing unit determines that the destination of the current data packet is Function 1 of HOST2 based on the internal flow forwarding engine logic. The data packet is then sent to the driver of Function 1 of HOST2 and then decapsulated by the protocol stack and sent to the user's upper-layer application. This process is repeated in this way to achieve internal forwarding. 2) External forwarding: The destination is an external device. Function 1 of HOST1 sends data to the external device. When the traffic comes out of Function 1 of HOS1, it is encapsulated by the protocol stack and then driven to the data processing unit. The data processing unit determines that the destination of the current data packet is the external device of the data processing unit based on the internal stream forwarding engine logic, so the data packet will be sent directly to the Ethernet port and then to the external device.

Images