A resource pooling management system for earth observation space cloud services

Abstract

The present invention provides a resource pooling management system for earth observation space cloud services, including a ground management terminal and an on-board management terminal. The ground management terminal is mainly responsible for the overall management of the resources of the entire constellation, and the on-board management terminal is mainly responsible for the resource pooling and management of multiple computing boards on a single satellite; the system can perform resource usage status statistics and generate an overall resource view. The present invention manages general computing resources and special computing resources in separate pools, manages and schedules resources on the basis of fully considering the satellite motion status and communication status, and can configure a complete C / S architecture in combination with ground management.

Description

Technical Field

[0001] The present invention belongs to the technical field of cloud computing, and in particular relates to a resource pooling management system for earth observation space cloud services. Background Art

[0002] Cloud computing is a technology that can provide scalable, distributed, and inexpensive computing power. Based on virtualization technology, cloud providers virtualize single server hardware into multiple virtual machines or containers to provide services to users. However, with the expansion of cloud computing business scale, traditional rack-mounted servers have begun to show fatigue when dealing with large-scale business models. When running a certain resource-intensive application, the efficiency is not high enough and the density is affected. At the same time, there is also a problem of life cycle asynchrony between different types of resources. For this reason, cloud providers usually customize different models of servers to run different types of tasks. However, with the increase in server types, not only does it bring difficulties to server resource management, but it also adds a lot of difficulties and costs to server operations. Therefore, many companies led by Intel and Google have begun to study new server architectures, among which server resource pooling technology has become the mainstream. Due to the small granularity of CPU, memory, and hard disk resources of a single server, resource configuration holes often appear in virtualization applications, either the number of CPU cores is not fully allocated, or the memory is not fully allocated, and more often there is a large amount of idle hard disk capacity. Resource pooling can better solve the above problems. Resource pooling provides virtualization capabilities at the hardware level. After resource pooling, virtualization and cloud server software can continue to run on the created logical servers, and virtual machines or containers can be created for users to use. Resource pooling can further reduce the resource configuration holes generated by cloud services and further improve the utilization of hardware resources. In addition to the advantages of flexible and elastic resource deployment and improved resource utilization, it can also more effectively improve the server's fault repair capabilities and improve server operation efficiency.

[0003] To ensure reliability, satellites generally use redundant designs, that is, multiple identical computing boards are installed to work simultaneously. Therefore, for a single satellite, how to pool its resources and manage them in a unified manner presents the following challenges:

[0004] 1. Current technologies are mainly oriented towards general computing resources such as CPU, memory, etc. However, for satellite-borne computing resources, their computing power is generally weak due to power consumption and other conditions. Therefore, satellites are generally equipped with specific chips such as FPGA to perform specific tasks. However, FPGA cannot perform general computing due to its characteristics, so it is difficult to pool it to support general tasks.

[0005] 2. The current technology is mainly aimed at data centers, whose servers are relatively stable. However, due to the lack of heat dissipation environment in space, a single computing board is difficult to work for a long time, so relatively frequent switching is required. Therefore, at a certain moment, the resources on different computing boards should be treated differently, which is not supported by current technology.

[0006] For the entire constellation, how to comprehensively manage resources presents the following challenges:

[0007] 1. Current resource management technology is mainly aimed at data centers. The positions and network status of each server are stable, but the relative positions of satellites in the constellation are constantly changing, and the network status is constantly changing. Therefore, when managing resources, it is necessary to fully consider the characteristics of satellites, which is not supported by existing technologies.

[0008] 2. Due to the limitations of onboard equipment performance and dynamic spatiotemporal connectivity, the intersatellite resource management system cannot be managed by a central server, and it is even more impossible to implement a C / S architecture between satellites.

[0009] From the above, it can be seen that the current resource pooling and management methods have certain problems when deployed on satellites, which are mainly reflected in three aspects. First, most of the chips on the satellite are dedicated chips, which are difficult to pool to provide general computing capabilities; second, the satellite is constantly moving and the network status is unstable; third, there is no center, and the C / S architecture cannot be deployed on the satellite. Summary of the invention

[0010] In view of the existing technical problems, the present invention provides a resource pooling management system for earth observation space cloud services. The present invention manages general computing resources and special computing resources in separate pools, manages and schedules resources on the basis of fully considering the satellite motion status and communication status, and can configure a complete C / S architecture in combination with the ground.

[0011] In order to achieve the above object, the technical solution adopted by the present invention is:

[0012] A resource pooling management system for earth observation space cloud services, comprising a ground management end and an on-board management end, wherein the ground management end is responsible for overall management of resources of the entire constellation, and the on-board management end is responsible for resource pooling and management of multiple computing boards on a single satellite; the ground management end comprises an ephemeris calculation module, a communication test module and a constellation management module, and the on-board management end comprises a general computing resource pooling management module, a special computing resource pooling management module and a single satellite management module;

[0013] The ephemeris calculation module is used to calculate the topology of the constellation based on the ephemeris to obtain the position and speed information of each satellite and send it to the constellation management module;

[0014] The communication test module is used to collect the bandwidth and latency of mutual access between the entire constellation topology and send it to the constellation management module;

[0015] The general computing resource pooling management module is used to pool the general computing resources on each computing board on the satellite and regularly report the usage of general computing resources to the single-satellite management module;

[0016] The dedicated computing resource pooling management module is used to format and describe the dedicated computing resources on each computing board and manage them in a pooled manner, and regularly report the usage of dedicated computing resources to the single-star management module;

[0017] The single-satellite management module is used to regularly package the usage of general computing resources and special computing resources, add timestamps, and then send them to the constellation management module;

[0018] The constellation management module is used to collect the resource usage of each satellite, unify the timestamp and generate the global resource usage; collect the information from the ephemeris calculation module, generate the position and velocity status for each satellite, and unify the timestamp to obtain the global satellite status; collect the information from the communication test module to generate the delay matrix between satellites; finally, unify the timestamp of the global resource usage, global satellite status and delay matrix to generate a dynamic panoramic resource view of the constellation.

[0019] The beneficial effects of the present invention are:

[0020] 1. The present invention manages general computing resources and special computing resources in separate pools, and can manage and schedule resources on the basis of fully considering the satellite motion status and communication status.

[0021] 2. The present invention can be combined with ground management to configure a complete C / S architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the overall structure of the system of the present invention;

[0023] Figure 2 Detailed structural diagram of the system of the present invention;

[0024] Figure 3 A flow chart showing resource usage status statistics in the method of the present invention;

[0025] Figure 4 A flowchart generated for the overall view of resources in the method of the present invention. DETAILED DESCRIPTION

[0026] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0027] A resource pooling management system for earth observation space cloud services, such as Figure 1As shown, the system is divided into two parts: the ground management end and the on-board management end. The ground management end is mainly responsible for the overall management of the resources of the entire constellation, and the on-board management end is mainly responsible for the resource pooling and management of multiple computing boards on a single satellite.

[0028] like Figure 2 As shown in the figure, the ground management end is divided into the ephemeris calculation module, the communication test module and the constellation management module. The onboard management end is divided into the general computing resource pooling management module, the dedicated computing resource pooling management module and the single satellite management module. The functions of each module of the ground management end are as follows:

[0029] The ephemeris calculation module is responsible for calculating the constellation topology based on the ephemeris to obtain the position, speed and other information of each star. The communication test module is responsible for collecting the bandwidth and delay of mutual access between the entire constellation topology. The constellation management module is responsible for collecting the resource information of each satellite and generating a panoramic resource view of the constellation in combination with the topology and network.

[0030] The functions of each module of the onboard management terminal are as follows:

[0031] The general computing resource pooling management module is responsible for pooling the general computing resources (such as CPU and memory) on each computing board on the satellite and regularly reporting the resource usage. The dedicated computing resource pooling management module is responsible for formatting and describing the dedicated computing resources (such as FPGA) on each computing board and managing them in a pool, and regularly reporting the resource usage. The single-satellite management module is responsible for regularly collecting the usage of general resources and dedicated resources and reporting to the ground.

[0032] The system can realize resource usage status statistics and generate an overall resource view.

[0033] like Figure 3 As shown, the specific method of resource usage status statistics is as follows:

[0034] Step 101: The general computing resource pooling management module pools the general resources on each computing board, which is recorded as:

[0035] R = {CPU, memory}

[0036] Among them, CPU represents CPU resources, and memory represents memory resources.

[0037] Then, the current general resource usage is sent to the single-star management module regularly.

[0038] Step 102, the dedicated computing resource pool management module formats and describes the dedicated resources on each computing board, and its function description is as follows:

[0039] F P ={UID,des}

[0040] Among them, uid is the identification number of the computing board, and des is the description information of the dedicated resources on the computing board.

[0041] Pool the dedicated computing resources of all computing boards and record them as:

[0042]

[0043] The subscripts 1, 2, ..., n represent the computing board numbers.

[0044] Then, the current dedicated resource usage is sent to the single-star management module regularly.

[0045] Step 103: The single-satellite management module periodically packages the usage of general computing resources and special computing resources, adds a timestamp, and sends the package to the constellation management module.

[0046] S = {T, R, F}

[0047] Among them, T is a unified timestamp.

[0048] like Figure 4 As shown in the figure, the specific method of generating the overall resource view is:

[0049] Step 201: The constellation management module collects the resource usage information sent by each satellite, unifies the timestamps and generates the global resource usage information, which is recorded as:

[0050] S={T,S1,S2...S n}

[0051] Among them, the subscripts 1, 2, ..., n represent the satellite numbers, and T is a unified timestamp.

[0052] Step 202: The constellation management module collects information from the ephemeris calculation module and generates a position and velocity state for each star, which is recorded as:

[0053] P i ={T,L i , V i}

[0054] Among them, the subscript i represents the satellite number, T is the timestamp, L is the position state, and V is the velocity state.

[0055] After unifying the timestamp, it is recorded as:

[0056] P={T,P1,P2...P n}

[0057] Step 203: The constellation management module collects information from the communication test module and generates a delay matrix between satellites, which is recorded as:

[0058]

[0059] Step 204: The constellation management module unifies the timestamps of the S, P, and N generated above to generate a dynamic panoramic resource view of the constellation.

[0060] In summary, the present invention manages general computing resources and special computing resources in separate pools, manages and schedules resources on the basis of fully considering the satellite motion status and communication status, and can configure a complete C / S architecture in combination with ground management.

Claims

1. A resource pooling management system for earth observation space cloud services, characterized in that: It includes a ground management end and an on-board management end, the ground management end is responsible for overall management of the resources of the entire constellation, and the on-board management end is responsible for resource pooling and management of multiple computing boards on a single satellite; the ground management end includes an ephemeris calculation module, a communication test module and a constellation management module, and the on-board management end includes a general computing resource pooling management module, a special computing resource pooling management module and a single satellite management module; The ephemeris calculation module is used to calculate the topology of the constellation based on the ephemeris to obtain the position and speed information of each satellite and send it to the constellation management module; The communication test module is used to collect the bandwidth and latency of mutual access between the entire constellation topology and send it to the constellation management module; The general computing resource pooling management module is used to pool the general computing resources on each computing board on the satellite and regularly report the usage of general computing resources to the single-satellite management module; The dedicated computing resource pooling management module is used to format and describe the dedicated computing resources on each computing board and manage them in a pooled manner, and regularly report the usage of dedicated computing resources to the single-star management module; The single-satellite management module is used to regularly package the usage of general computing resources and special computing resources, add timestamps, and then send them to the constellation management module; The constellation management module is used to collect the resource usage of each satellite and generate the global resource usage after unifying the timestamp; Collect information from the ephemeris calculation module, generate position and velocity status for each satellite, unify the timestamp, and obtain the global satellite status; Collect information from the communication test module and generate a delay matrix between satellites. Finally, unify the global resource usage, global satellite status, and delay matrix with a unified timestamp to generate a dynamic panoramic resource view of the constellation.

Images