Communication system, control program and control method

Abstract

To maintain a base station function even upon a node failure.SOLUTION: A communication system 100 including a base station having a Composable Disaggregated Infrastructure(CDI) configuration, includes a control unit 1 that stores a measurement result 142 of a latency between all nodes 21 and all accelerators 231 included in the communication system 100, and when a node failure has occurred, identifies a standby node 21 satisfying a latency requirement 145 from a standby node group on the basis of positional information for the accelerator connected to a failure node 21, and switches the failure node 21 to the standby node 21.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a communication system, a control program, and a control method. [Background technology]

[0002] In mobile communications networks, efforts are underway to virtualize base stations, which replace dedicated base station equipment that controls antennas with general-purpose servers and signal processing accelerators.

[0003] Additionally, efforts are underway to develop a Centralized Radio Access Network (C-RAN), which consolidates base station equipment in station buildings in each region, such as cities and towns.

[0004] By combining virtualized base stations with C-RAN, it is expected that station buildings will be used as data centers. By using the servers in the buildings not only as base stations but also as Multi-access Edge Computing (MEC), hardware resources can be utilized more effectively.

[0005] Composable Disaggregated Infrastructure (CDI) technology is being developed in the server industry to further improve the efficiency of hardware resource utilization. CDI is a technology that divides and reconstructs hardware resources such as the central processing unit (CPU), memory, storage, network controller, and accelerator, which are the components of a server.

[0006] By reconfiguring hardware resources according to the server's intended use, it is expected that resource utilization efficiency and power efficiency will improve, especially in medium- to large-scale data centers. For example, in a mobile communications network, by allocating a signal processing accelerator to base stations and a graphics processing unit (GPU) to MEC, the equipment price and power consumption can be reduced compared to when both components are installed in each server.

[0007] In the server rack installed in the base station equipment, a component box equipped with components such as accelerators and GPUs is deployed outside a compute node (hereinafter referred to as a "node") equipped with general-purpose computing resources. The components can be connected to the nodes via fabric switches according to instructions from a resource manager.

[0008] The components allocated by the resource manager can be changed depending on the application, such as a base station or MEC, making it possible to flexibly create servers for each application.

[0009] CDI can also consist of multiple server racks. For example, by interconnecting the fabric switches between the server racks, it is possible to allocate components across server racks. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] International Publication No. 2014 / 118938 [Patent Document 2] JP 2008-107896 A Summary of the Invention [Problem to be solved by the invention]

[0011] In a 5G virtual base station configured with CDI, a combination of a node and an accelerator may be prepared as a standby node. In such a case, due to timing requirements and the number of antennas accommodated by the base station (in other words, the number of cells), not all combinations of nodes and accelerators may function as standby nodes. 5G virtual base stations have strict timing requirements (approximately 100 ns to several us, depending on the communication function used), and the requirements may not be met depending on the distance between the combined node and accelerator.

[0012] Because base station servers use an Operating System (OS) that is suitable for real-time processing, converting the MEC server close to the accelerator into a base station server is also not an option, as it would require replacing the OS. Replacing the OS takes anywhere from a few tens of minutes to a few hours, during which there is a risk of the quality of the communication service declining or being stopped. In particular, any decline in communication quality or stoppage for longer than a certain period of time may trigger an obligation to report to the competent authorities, so conversion that involves replacing the OS may be risky.

[0013] In this way, there is a risk that the resource manager may not be able to switch between standby nodes and select an accelerator that satisfies the timing requirements and the number of antennas (in other words, the number of cells) of the base station.

[0014] One aspect is to maintain base station functionality even in the event of a node failure. [Means for solving the problem]

[0015] In one aspect, a communication system includes a base station having a Composable Disaggregated Infrastructure (CDI) configuration, and stores measurement results of latency between all nodes and all accelerators included in the communication system. When a node failure occurs, a standby node that satisfies the latency requirements is identified from a group of standby nodes based on the latency and location information of the accelerator connected to the failed node, and the failed node is switched to the standby node. Effect of the Invention

[0016] In one aspect, the base station function can be maintained even in the event of a node failure. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a block diagram illustrating a configuration example of a virtualized base station system to which CDI is applied, as a related example. [Diagram 2] 2 is a block diagram illustrating a process when a node fails in the virtualized base station system shown in FIG. 1. FIG. [Diagram 3] 1 is a block diagram illustrating a configuration example of a virtualized base station system according to an embodiment. [Figure 4] 4 is a block diagram illustrating an example of a hardware configuration of a resource manager illustrated in FIG. 3. [Diagram 5] 4 is a diagram for explaining a process of searching for a standby node of a faulty rack in a resource table in the resource manager shown in FIG. 3. [Figure 6] 4 is a diagram for explaining a search process for a standby node in the failed rack shown in FIG. 3. [Figure 7] 4 is a diagram for explaining a search process for a standby node of another rack in a resource table and a latency table in the resource manager shown in FIG. 3. [Figure 8] 4 is a diagram for explaining a search process for a standby node in another rack shown in FIG. 3. FIG. [Figure 9] 4 is a diagram for explaining a process of searching for a standby node and a standby accelerator of another rack in a resource table and an accelerator table in the resource manager shown in FIG. 3. FIG. [Figure 10] 4 is a diagram for explaining a search process for a standby node and a standby accelerator in another rack shown in FIG. 3. FIG. [Figure 11] 4 is a diagram for explaining a process of searching for a standby node of another rack and a standby accelerator of another rack in the resource table, the latency table, and the accelerator table in the resource manager shown in FIG. 3. FIG. [Figure 12] 4 is a diagram for explaining a search process for a standby node in another rack and a standby accelerator in another rack shown in FIG. 3. FIG. [Figure 13] 11 is a flowchart illustrating a node switching process according to an embodiment. [Figure 14] 14 is a flowchart illustrating details of the process of switching step (1) in step S9 of FIG. 13. [Figure 15] 14 is a flowchart illustrating details of the process of switching step (2) in step S11 of FIG. 13. [Figure 16] 14 is a flowchart illustrating details of the process of switching step (3) in step S13 of FIG. 13. [Figure 17] 14 is a flowchart illustrating details of the process of switching step (4) in step S15 of FIG. 13. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] [A] Embodiment An embodiment will be described below with reference to the drawings. However, the embodiment described below is merely an example, and there is no intention to exclude the application of various modified examples and techniques not specified in the embodiment. In other words, the present embodiment can be implemented with various modifications within the scope of the gist of the embodiment. In addition, each figure does not intend to include only the components shown in the figure, but can include other components, etc.

[0019] [A-1] Related Examples FIG. 1 is a block diagram showing a schematic configuration example of a virtualized base station system 200 to which CDI is applied, as a related example.

[0020] The virtualized base station system 200 includes a resource manager 1, multiple (N in the example shown in FIG. 1; N is an integer equal to or greater than 2) server racks 2, and a network switch 3. The server racks 2 may be referred to as server racks #1 to #N. The virtualized base station system 200 is also connected to multiple (three in the example shown in FIG. 1) antennas (Radio Units; RUs) 4.

[0021] The network switch 3 routes communications between the server racks 2, and also routes communications between the virtualized base station system 200 and an external system.

[0022] Each server rack 2 includes a plurality of nodes 21 (four in the example shown in FIG. 1), fabric switches (SWs) 22, and component boxes 23.

[0023] Each node 21 includes a plurality of CPUs 211 .

[0024] The fabric SW22 routes communications between each node 21 and the component box 23, and routes communications between the server rack 2 itself and other server racks 2.

[0025] The component box 23 includes a plurality of accelerators 231 and a plurality of GPUs 232 .

[0026] 1, in a server rack #1, one node 21 and a plurality of accelerators 231 in a component box 23 function as a base station server 20. In addition, in a server rack #N, one node 21 and a plurality of GPUs 232 in a component box 23 function as an MEC server 30.

[0027] That is, the accelerator 231 functions as a resource for the base station server 20 , and the GPU 232 functions as a resource for the MEC server 30 .

[0028] The resource manager 1 manages resources in each server rack 2 , and allocates the resources in each server rack 2 to the base station server 20 or the MEC server 30 .

[0029] FIG. 2 is a block diagram for explaining a process when a node fails in the virtualized base station system 200 shown in FIG.

[0030] In the example indicated by reference symbol A1 in FIG. 2, one node 21 and a plurality of accelerators 231 in a component box 23 function as a base station server 20 in server racks #1, #2, and #N.

[0031] As shown by reference symbol A11, it is assumed that a failure occurs in the CPU 211 of the node 21 assigned to the base station server 20 in the server rack #N.

[0032] When a failure occurs in a node 21, the failed node 21 is switched to a standby node 21 (for example, one node 21 in the server rack #1 denoted by reference symbol A1).

[0033] In the example shown by symbol A2, one node 21 in server rack #1 is switched to the operational system, and the node 21 in server rack #1 and the accelerator 231 in server rack #N connected to the failed node 21 function as a new base station server 20 (NEW).

[0034] However, as the distance between node 21 and accelerator 231 increases, the timing requirements of the base station may not be met.

[0035] [A-2] Configuration example FIG. 3 is a block diagram illustrating a schematic configuration example of a virtualized base station system 100 in an embodiment.

[0036] The virtualized base station system 100 is an example of a communication system, and includes a resource manager 1 and multiple (N in the example shown in FIG. 1; N is an integer equal to or greater than 2) server racks 2. The server racks 2 may be referred to as server racks #1 to #N.

[0037] Each server rack 2 includes a plurality of nodes 21 (two in the example shown in FIG. 3), a fabric switch (SW) 22, and a component box 23.

[0038] The node 21 functions as a latency measurement unit 201 and a base station function notification unit 202 .

[0039] The latency measurement unit 201 measures the speed (in other words, latency) between the node 21 itself and all the accelerators 231 when the virtualized base station system 100 is started up for the first time.

[0040] The base station function notification unit 202 notifies the latency table creation unit 111 of the resource manager 1 of the measurement result of the latency measurement unit 201 .

[0041] The fabric SW22 routes communications between each node 21 and the component box 23.

[0042] The component box 23 includes a plurality of accelerators 231 .

[0043] The resource manager 1 manages resources in each server rack 2 , and allocates the resources in each server rack 2 to the base station server 20 or the MEC server 30 .

[0044] The resource manager 1 stores the measurement results of the speed (in other words, latency) between all nodes 21 and all accelerators 231 when the virtualized base station system 100 is initially launched within the 5G virtual base station.

[0045] The resource manager 1 functions as a latency table creation unit 111, an accelerator table creation unit 112, and a base station requirement value setting unit 113. The resource manager 1 also stores a resource table 141, a latency table 142, an accelerator table 143, and a base station requirement 144. The base station requirement 144 includes a latency requirement value 145 and a number of cells requirement value 146.

[0046] The latency table creation unit 111 creates a latency table 142 based on the latency measurement results by the latency measurement unit 201 in each node 21 of each server rack 2 .

[0047] The accelerator performance setting unit 114 sets the performance (number of cells) of the accelerator 231 mounted in the component box 23 in the cell performance value 147 .

[0048] The resource manager 1 generates the cell performance values ​​147 by obtaining them from the operating software (including management software such as a controller).

[0049] The accelerator table creation unit 112 acquires model name information of each accelerator 231 provided in the component box 23 from each server rack 2 , and creates the accelerator table 143 by referring to the cell performance value 147 .

[0050] The base station requirement value setting unit 113 sets the base station requirement 144 including a latency requirement value 145 and a cell number requirement value 146 based on the required performance of the base station server 20 .

[0051] The resource manager 1 generates base station requirements 144 for the 5G virtual base station by obtaining a latency requirement value 145 and a cell number requirement value 146 from operation software (including management software such as a controller).

[0052] When a node failure occurs, the resource manager 1 searches for and switches to a standby node 21 that satisfies the timing requirements of another rack 2 based on the measured speed, timing requirements, and the installation information of the accelerator 231 that was connected to the failed node.

[0053] If there is no standby node 21 in the other rack 2 that satisfies the timing requirement, the resource manager 1 searches for a combination of the standby node 21 and the accelerator 231 in the other rack 2 that satisfies the timing requirement and the cell number requirement, and switches the combination.

[0054] This will enable CDI to improve the efficiency of hardware resource utilization and power efficiency, while also making it possible to continue maintaining base station functions even in the event of a node failure.

[0055] FIG. 4 is a block diagram showing a schematic example of a hardware configuration of the resource manager 1 shown in FIG.

[0056] As shown in FIG. 4, the resource manager 1 includes a CPU 11, a memory 12, a display control device 13, a storage device 14, an input interface (IF) 15, an external recording medium processing device 16, and a communication IF 17.

[0057] The memory 12 is an example of a storage unit, and is illustratively a Read Only Memory (ROM) and a Random Access Memory (RAM). Programs such as a Basic Input / Output System (BIOS) may be written in the ROM of the memory 12. The software programs in the memory 12 may be appropriately read and executed by the CPU 11. The RAM of the memory 12 may be used as a temporary storage memory or a working memory.

[0058] The display control device 13 is connected to a display device 131 and controls the display device 131. The display device 131 is a liquid crystal display, an organic light-emitting diode (OLED) display, a cathode ray tube (CRT), an electronic paper display, or the like, and displays various information to an operator, etc. The display device 131 may be combined with an input device, for example, a touch panel. The display device 131 displays various information to a user of the resource manager 1.

[0059] The storage device 14 is a storage device with high IO performance, and may be, for example, a dynamic random access memory (DRAM), a solid state drive (SSD), a storage class memory (SCM), or a hard disk drive (HDD). The storage device 14 stores the measurement results of the latency.

[0060] The input IF 15 may be connected to input devices such as a mouse 151 and a keyboard 152, and may control the input devices such as the mouse 151 and the keyboard 152. The mouse 151 and the keyboard 152 are examples of input devices, and an operator performs various input operations via these input devices.

[0061] The external recording medium processing device 16 is configured so that the recording medium 160 can be attached thereto. The external recording medium processing device 16 is configured so that, when the recording medium 160 is attached thereto, the information recorded on the recording medium 160 can be read. In this example, the recording medium 160 is portable. For example, the recording medium 160 is a non-transitory recording medium such as a flexible disk, an optical disk, a magnetic disk, a magneto-optical disk, or a semiconductor memory.

[0062] The communication IF 17 is an interface for enabling communication with an external device.

[0063] The CPU 11 is an example of a processor, and is a processing device that performs various controls and calculations. The CPU 11 realizes various functions by executing an Operating System (OS) and programs loaded into the memory 12. The CPU 11 may be a multiprocessor including multiple CPUs, a multicore processor having multiple CPU cores, or a configuration having multiple multicore processors.

[0064] The device for controlling the operation of the entire resource manager 1 is not limited to the CPU 11, and may be, for example, any one of MPU, DSP, ASIC, PLD, and FPGA. The device for controlling the operation of the entire resource manager 1 may be a combination of two or more of CPU, MPU, DSP, ASIC, PLD, and FPGA. Note that MPU is an abbreviation for Micro Processing Unit, DSP is an abbreviation for Digital Signal Processor, and ASIC is an abbreviation for Application Specific Integrated Circuit. Also, PLD is an abbreviation for Programmable Logic Device, and FPGA is an abbreviation for Field Programmable Gate Array.

[0065] The resource manager 1 may execute the node switching process when a node failure occurs through the following switching steps (1) to (4).

[0066] (1) The resource manager 1 searches for a standby node 21 in the failed rack 2. When a node 21 fails, the resource manager 1 searches for an allocatable standby node 21 from among the standby nodes 21 in the same server rack 2. Note that delay time does not need to be taken into consideration at this time.

[0067] (2) The resource manager 1 searches for a standby node 21 in another rack 2 other than the failed rack 2. If the standby node 21 is not found in the switching step (1), the resource manager 1 searches for a standby node 21 in another rack 2 that can be connected to the accelerator 231 in the failed rack 2 and satisfies the allowable delay time of the currently running program.

[0068] (3) The resource manager 1 searches for a standby node 21 and a standby accelerator 231 in another rack 2. If the resource manager 1 cannot find the standby node 21 even in the switching step (2), the resource manager 1 searches for a standby accelerator 231 and a standby node 21 in another rack 2 that satisfy the number of antennas (in other words, the number of cells) required by the currently running program.

[0069] (4) The resource manager 1 searches for the standby node 21 in the other rack 2 and the standby accelerator 231 in the other rack 2. If the combination of the standby node 21 and the accelerator 231 is not found even in the switching step (3), the resource manager 1 searches for the standby node 21 in the other rack 2 whose delay time with the standby accelerator 231 in the other rack 2 that satisfies the required number of antennas (in other words, the number of cells) of the currently running program satisfies the allowable delay time of the currently running program.

[0070] Fig. 5 is a diagram for explaining a search process for a standby node in the faulty rack 2 in the resource table 141 in the resource manager 1 shown in Fig. 3. Fig. 6 is a diagram for explaining a search process for a standby node in the faulty rack 2 shown in Fig. 3.

[0071] In the switching step (1), if a failure (see symbol B3 in Figure 6) occurs in the CPU 211 installed in the node 21 in the server rack #1, the resource manager 1 searches the resource table 141 to see if there is a standby node 21 among the nodes in the server rack #1.

[0072] In the example shown in Figure 5, the resource manager 1 refers to the resource table 141 to determine whether node #1 of server rack #1 is in use (see symbol B1 in Figure 5) and whether node #3 of server rack #1 is in use (see symbol B2 in Figure 5).

[0073] If there is an unused node 21 in the failed rack 2, the resource manager 1 disconnects the accelerators (ACC) #3 and #4 that are connected to the failed node 21, node #2. The resource manager 1 performs connection settings between the accelerators #3 and #4 and node #3 (see symbol B4 in FIG. 6), which is a standby node. Then, the resource manager 1 starts up node #3 and starts operation.

[0074] On the other hand, if there is no unused node, the process proceeds to the next switching step (2).

[0075] Fig. 7 is a diagram for explaining a search process for a standby node 21 in another rack 2 in the resource table 141 and the latency table 142 in the resource manager 1 shown in Fig. 3. Fig. 8 is a diagram for explaining a search process for a standby node 21 in another rack 2 shown in Fig. 3.

[0076] The resource manager 1 refers to the latency table 142 and extracts one or more server racks 2 that have a connection combination with the component BOX#1 that results in a latency value smaller than the latency requirement value (for example, 1500 ns) (see symbol C1 in FIG. 7).

[0077] The resource manager 1 refers to the resource table 141 to search for the presence or absence of a standby node 21 in the extracted server rack 2 .

[0078] When there are multiple standby nodes 21, the resource manager 1 selects the standby node 21 in the server rack 2 having the smallest value in the latency table 142 (see symbol C3 in FIG. 7).

[0079] If there is a standby node 21, the resource manager 1 disconnects the accelerators #3 and #4 of the server rack #1 that are set to be connected to node #2 of the server rack #1. The resource manager 1 performs connection settings between the selected standby node #3 (see symbol C2 in FIG. 7 and symbol D1 in FIG. 8) and the accelerators #3 and #4 of the server rack #1. Then, the resource manager 1 starts up the unused node 21 and starts operation.

[0080] On the other hand, if there is no standby node 21, the process proceeds to the next switching step (3).

[0081] Fig. 9 is a diagram for explaining a search process for the standby node 21 and the standby accelerator 231 in another rack 2 in the resource table 141 and the accelerator table 143 in the resource manager 1 shown in Fig. 3. Fig. 10 is a diagram for explaining a search process for the standby node 21 and the standby accelerator 231 in another rack 2 shown in Fig. 3.

[0082] The resource manager 1 searches the resource table 141 for the standby node 21 (see symbol E1 in FIG. 9).

[0083] The resource manager 1 extracts one or more standby accelerators 231 that are in the same rack 2 as the standby node 21 (see symbol E2 in FIG. 9).

[0084] The resource manager 1 obtains the number of cells of the accelerator 231 extracted from the accelerator table 143 (see symbol E3 in FIG. 9).

[0085] The resource manager 1 judges whether the total number of cells obtained is equal to or greater than the cell number requirement value 146 (for example, 12 cells; see symbol E4 in FIG. 9) of the base station requirement 144.

[0086] If the total number of acquired cells is equal to or greater than the cell number requirement value 146, the resource manager 1 disconnects accelerators #3 and #4 of server rack #1 that are set to be connected to node #2 of server rack #1. The resource manager 1 stops the failed node #2 of server rack #1, and connects accelerators #5 and #6 of server rack #7 to node #3 (see symbol F1 in FIG. 10), which is a standby node in server rack #7. Then, the resource manager 1 starts up node #3 of server rack #7.

[0087] On the other hand, if the total number of cells obtained is less than the cell number requirement value 146, the process proceeds to the next switching step (4).

[0088] Fig. 11 is a diagram for explaining a search process for the standby node 21 in the other rack 2 and the standby accelerator 231 in the other rack 2 in the resource table 141, the latency table 142, and the accelerator table 143 in the resource manager 1 shown in Fig. 3. Fig. 12 is a diagram for explaining a search process for the standby node 21 in the other rack 2 and the standby accelerator 231 in the other rack 2 shown in Fig. 3.

[0089] The resource manager 1 refers to the latency table 142, and extracts one or more component boxes 23 having a latency smaller than the latency requirement value 145 (e.g., 1500 ns) of the base station requirement 144, based on the rack 2 mounting the standby node 21 in the previous switching step (3). Here, the component box #8 connected to the node 21 in the server rack #7 is extracted (see symbol G1 in FIG. 11).

[0090] The resource manager 1 refers to the resource table 141, and extracts one or more standby accelerators 231 of the extracted component box 23 (see symbol G2 in FIG. 11).

[0091] The resource manager 1 refers to the accelerator table 143 and acquires the number of cells of the extracted standby accelerator 231 (see symbol G3 in FIG. 11).

[0092] The resource manager 1 judges whether the total of the acquired cell numbers is equal to the cell number requirement value 146 (for example, 12 cells; see symbol G4 in FIG. 11) of the base station requirement 144.

[0093] If the acquired number of cells is equal to or greater than the cell number requirement value 146, the resource manager 1 disconnects accelerators #3 and #4 of server rack #1 that are set to be connected to node #2 of server rack #1, and stops node #2 of server rack #1. The resource manager 1 connects accelerator #6 of server rack #7 to node #3 of server rack #7 (see symbol H1 in FIG. 12), and connects accelerator #2 of server rack #8 to node #3 of server rack #7 (see symbol H1 in FIG. 12). Then, the resource manager 1 starts node #3 of server rack #7.

[0094] On the other hand, if the total number of cells obtained is less than the number of cells requirement value 146, the process returns to the switching step (1) and the accelerator 231 is searched for.

[0095] [A-3] Example of operation The node switching process in this embodiment will be described with reference to the flowchart (steps S1 to S17) shown in FIG.

[0096] When the virtualized base station system 100 is started up, the resource manager 1 instructs each node 21 to measure latency (step S1).

[0097] The latency measurement unit 201 measures the latency in each node 21 and notifies the resource manager 1 of the measurement result (step S2).

[0098] The latency table creating unit 111 of the resource manager 1 creates a latency table 142 from the latency measurement results notified from each node 21 (step S3).

[0099] The resource manager 1 sets the cell performance value 147 in the accelerator performance setting unit 114 (step S4).

[0100] The resource manager 1 acquires information on the accelerators 231 mounted in each component box 23 from the resource table 141 (step S5).

[0101] The accelerator table creating unit 112 of the resource manager 1 creates the accelerator table 143 by referring to the cell performance value 147 based on the accelerator information (step S6).

[0102] The base station requirement value setting unit 113 of the resource manager 1 sets the latency requirement value 145 and the number of cells requirement value 146, which are the base station requirements 144 (step S7).

[0103] The virtualized base station system 100 is put into operation (step S8).

[0104] The resource manager 1 determines whether a node failure has been detected (step S9).

[0105] If a node failure has not been detected (see the No route from step S9), the process returns to step S8.

[0106] On the other hand, if a node failure is detected (see the Yes route in step S9), the resource manager 1 executes the process of the switching step (1) (step S10). The details of the process of the switching step (1) will be described later with reference to FIG.

[0107] The resource manager 1 judges whether the failure has been restored (step S11).

[0108] If the failure has been restored (see the Yes route from step S11), the process returns to step S7.

[0109] On the other hand, if the failure has not been restored (see the No route in step S11), the resource manager 1 executes the process of the switching step (2) (step S12). The details of the process of the switching step (2) will be described later with reference to FIG.

[0110] The resource manager 1 judges whether the failure has been restored (step S13).

[0111] If the failure has been restored (see the Yes route from step S13), the process returns to step S7.

[0112] On the other hand, if the failure has not been restored (see the No route in step S13), the resource manager 1 executes the process of the switching step (3) (step S14). The details of the process of the switching step (3) will be described later with reference to FIG.

[0113] The resource manager 1 judges whether the failure has been restored (step S15).

[0114] If the failure has been restored (see the Yes route from step S15), the process returns to step S7.

[0115] On the other hand, if the failure has not been restored (see the No route in step S15), the resource manager 1 executes the process of the switching step (4) (step S16). The details of the process of the switching step (5) will be described later with reference to FIG.

[0116] The resource manager 1 judges whether the failure has been restored (step S17).

[0117] If the failure has been restored (see the Yes route from step S17), the process returns to step S8.

[0118] On the other hand, if the failure has not been restored (see the No route in step S17), the resource manager 1 notifies the operation manager that switching is not possible (step S18). The notification of switching is not possible may be performed using the display device 131 shown in Fig. 4. Then, the node switching process ends.

[0119] Next, details of the process of the switching step (1) in step S10 in FIG. 13 will be described with reference to the flowchart (steps S21 to S25) shown in FIG.

[0120] The resource manager 1 refers to the resource table 141 (step S21).

[0121] The resource manager 1 determines whether or not there is a standby node 21 in the server rack 2 in which the node failure has occurred (step S22).

[0122] If there is no standby node 21 in the server rack 2 in which the node failure has occurred (see the No route in step S22), the process of the switching step (1) ends.

[0123] On the other hand, if there is a standby node 21 in the server rack 2 where the node failure has occurred (see the Yes route in step S22), the resource manager 1 disconnects the accelerator 231 of the component BOX 23 connected to the failed node (step S23).

[0124] The resource manager 1 connects the accelerator 231 of the component box 23 that has been disconnected from the failed node 21 to the standby node 21 (step S24).

[0125] The resource manager 1 switches the standby node 21 to the operating node 21 to recover from the failure (step S25), and the process of the switching step (1) is then completed.

[0126] Next, details of the process of the switching step (2) in step S12 of FIG. 13 will be described with reference to the flowchart (steps S31 to S37) shown in FIG.

[0127] The resource manager 1 refers to the latency table 142 (step S31).

[0128] The resource manager 1 extracts server racks 2 having a smaller latency requirement value than the latency requirement value 145 of the base station requirement 144 from among the combinations of the component box 23 connected to the failed node 21 and other server racks 2 (step S32).

[0129] The resource manager 1 determines whether or not there is a standby node 21 in the extracted server rack 2 (step S34).

[0130] If there is no standby node 21 in the extracted server rack 2 (see the No route in step S34), the process of the switching step (2) ends.

[0131] On the other hand, if the extracted server rack 2 contains a standby node 21 (see the Yes route in step S34), the resource manager 1 disconnects the accelerator 231 of the component BOX 23 connected to the failed node 21 (step S35).

[0132] The resource manager 1 connects the accelerator 231 of the component box 23 that has been disconnected from the failed node 21 to the standby node 21 (step S36).

[0133] The resource manager 1 switches the standby node 21 to the operating node 21 to recover from the failure (step S37), and the process of the switching step (2) is then completed.

[0134] Next, the details of the process of the switching step (3) in step S14 of FIG. 13 will be described with reference to the flowchart (steps S41 to S52) shown in FIG.

[0135] The resource manager 1 refers to the resource table 141 (step S41).

[0136] The resource manager 1 extracts server racks 2 having a smaller latency requirement value than the latency requirement value 145 of the base station requirement 144 from among the combinations of the component box 23 connected to the failed node 21 and other server racks 2 (step S42).

[0137] The resource manager 1 determines whether the server rack 2 in which the standby node 21 is located has been detected (step S43).

[0138] If the server rack 2 in which the standby node 21 is located cannot be detected (see the No route in step S43), the resource manager 1 notifies the operation manager that switching is not possible (step S44). The notification of switching is not possible may be performed using the display device 131 shown in Fig. 4. Then, the processing of the switching step (3) (in other words, the node switching processing) ends.

[0139] On the other hand, if a server rack 2 with a standby node 21 is detected (see the Yes route in step S43), the resource manager 1 searches for a standby accelerator 231 mounted in a component box 23 in the searched server rack 2 (step S45).

[0140] The resource manager 1 determines whether the standby accelerator 231 has been detected (step S46).

[0141] If the standby accelerator 231 cannot be detected (see the No route in step S46), the process of the switching step (3) ends.

[0142] On the other hand, if the standby accelerator 231 is detected (see the Yes route of step S46), the resource manager 1 refers to the accelerator table 143 (step S47).

[0143] The resource manager 1 acquires the cell number information of the standby accelerator 231 found in the resource table 141 (step S48).

[0144] The resource manager 1 judges whether the total of the acquired cell number information of the standby accelerators 231 is equal to or greater than the cell number requirement value 146 of the base station requirement 144 (step S49).

[0145] If the total of the acquired cell number information of the standby accelerators 231 is not equal to or greater than the cell number requirement value 146 of the base station requirement 144 (see the No route of step S49), the process of the switching step (3) ends.

[0146] On the other hand, if the sum of the acquired cell count information of the standby accelerators 231 is greater than or equal to the cell count requirement value 146 of the base station requirement 144 (see the Yes route in step S49), the resource manager 1 disconnects the accelerators 231 of the failed node 21 and the component BOX 23 connected to the failed node 21 (step S50).

[0147] The resource manager 1 connects the standby node 21 of the server rack 2 in which the standby node 21 is located to the standby accelerator 231 of the component box 23 (step S51).

[0148] The resource manager 1 switches the standby node 21 to the operating node 21 to recover from the failure (step S52), and the process of the switching step (3) is then completed.

[0149] Next, the details of the process of the switching step (4) in step S16 in FIG. 13 will be described with reference to the flowchart (steps S61 to S71) shown in FIG.

[0150] The resource manager 1 refers to the latency table 142 (step S61).

[0151] The resource manager 1 extracts a server rack 2 having a smaller latency requirement value 145 than the base station requirement 144 from the combination of the component box 23 in the server rack 2 of the standby node 21 searched for in the switching step (3) and another server rack 2 (step S62).

[0152] The resource manager 1 refers to the resource table 141 (step S63).

[0153] The resource manager 1 searches for a standby accelerator 231 from the extracted server rack 2 (step S64).

[0154] The resource manager 1 determines whether the standby accelerator 231 has been detected (step S65).

[0155] If the standby accelerator 231 cannot be detected (see the No route in step S65), the process returns to the switching step (3).

[0156] On the other hand, if the standby accelerator 231 is detected (see the Yes route of step S65), the resource manager 1 refers to the accelerator table 143 (step S66).

[0157] The resource manager 1 judges whether the total of the cell number information of the detected standby accelerators 231 is equal to or greater than the cell number requirement value 146 of the base station requirement 144 (step S67).

[0158] If the sum of the cell count information of the detected waiting accelerators 231 is less than the cell count requirement value 146 of the base station requirement 144 (see the No route of step S67), the resource manager 1 determines whether all resources in the virtualized base station system 100 have been confirmed (step S68).

[0159] If all resources in the virtualized base station system 100 have not been confirmed (see the No route from step S68), the process returns to the switching step (3).

[0160] On the other hand, if all resources in the virtualized base station system 100 have been confirmed (see the Yes route from step S68), the process of the switching step (4) ends.

[0161] In step S67, if the sum of the cell count information of the detected standby accelerators 231 is equal to or greater than the cell count requirement value 146 of the base station requirement 144 (see the Yes route in step S67), the resource manager 1 disconnects the accelerators 231 of the failed node 21 and the component BOX 23 connected to the failed node 21 (step S69).

[0162] The resource manager 1 connects the standby node 21 of the server rack 2 in which the standby node 21 is located to the standby accelerator 231 of the component box 23 (step S70).

[0163] The resource manager 1 switches the standby node 21 to the operating node 21 to recover from the failure (step S71), and the process of the switching step (4) is then completed.

[0164] [B] Effect According to the communication system, the control program, and the control method in the above-described embodiments, for example, the following advantageous effects can be obtained.

[0165] The resource manager 1 stores the measurement results of the latency between all the nodes 21 and all the accelerators 231 included in the virtualized base station system 100. When a node failure occurs, the resource manager 1 identifies a standby node 21 that satisfies the latency requirement from a group of standby nodes based on the latency and the position information of the accelerator 231 connected to the failed node 21. The resource manager 1 switches the failed node 21 to the standby node 21.

[0166] This makes it possible to maintain base station functionality even in the event of a node failure. Specifically, it is possible to improve the efficiency of hardware resource utilization and power efficiency in CDI while continuing to maintain base station functionality even in the event of a node failure.

[0167] The resource manager 1 identifies a standby node 21 that satisfies the latency requirement as well as the cell number requirement from the standby node group based on the latency, the location information, and the number of cells that the accelerator 231 can accommodate.

[0168] As a result, even after node 21 is switched, it is possible to perform normal communication as a base station.

[0169] The resource manager 1 identifies the standby node 21 from the first standby node group provided in the same first server rack 2 as the failed node 21, as the location information.

[0170] This allows switching to a standby node 21 in the same server rack 2 as the failed node 21, thereby reducing the number of configuration changes to the virtualized base station system 100.

[0171] When the first group of standby nodes does not exist, the resource manager 1 identifies a standby node 21 from a second group of standby nodes that is provided in a second server rack 2 different from the first server rack 2 and whose latency with the accelerator 231 in the first server rack 2 satisfies the latency requirement.

[0172] This makes it possible to use a node 21 in a server rack 2 different from the node 21 in which the failure occurred.

[0173] When a second group of standby nodes does not exist, the resource manager 1 identifies a standby node 21 from a third group of standby nodes that is provided in a second server rack 2 and whose latency with the accelerator 231 in the second server rack 2 satisfies the latency requirement.

[0174] This makes it possible to use the node 21 and the accelerator 231 in a server rack 2 different from the node 21 in which the failure occurred.

[0175] When the third standby node group does not exist, the resource manager 1 identifies a standby node 21 from a fourth standby node group that is provided in a second server rack 2 and whose latency with an accelerator 231 in a third server rack 2 different from the first server rack 2 or the second server rack 2 satisfies the latency requirement.

[0176] This makes it possible to use a node 21 in a server rack 2 different from the failed node 21 and an accelerator 231 in a server rack 2 different from the node 21.

[0177] [C] Other The disclosed technology is not limited to the above-described embodiment, and various modifications can be made without departing from the spirit of the present embodiment. Each configuration and each process of the present embodiment can be selected as necessary, or can be combined as appropriate.

[0178] [D] Additional Notes The following supplementary notes are further disclosed regarding the above embodiment.

[0179] (Appendix 1) A communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A communication system comprising a control unit.

[0180] (Appendix 2) The control unit is Identifying the standby node that satisfies a cell number requirement in addition to the latency requirement from the standby node group based on a number of cells that the accelerator can accommodate in addition to the latency and the location information; 2. The communication system of claim 1.

[0181] (Appendix 3) The control unit is As the location information, the standby node is identified from a first standby node group provided in the same first server rack as the failed node. 3. A communication system according to claim 1 or 2.

[0182] (Appendix 4) The control unit is When the first group of standby nodes does not exist, the standby node is identified from a second group of standby nodes that is provided in a second server rack different from the first server rack and whose latency with the accelerator in the first server rack satisfies the latency requirement. 4. A communication system as described in claim 3.

[0183] (Appendix 5) The control unit is When the second group of standby nodes does not exist, identifying the standby node from a third group of standby nodes that is provided in the second server rack and whose latency with the accelerator in the second server rack satisfies the latency requirement. 5. The communication system of claim 4.

[0184] (Appendix 6) The control unit is When the third group of standby nodes does not exist, the standby node is identified from a fourth group of standby nodes that is provided in the second server rack and whose latency with the accelerator in a third server rack different from the first server rack or the second server rack satisfies the latency requirement. 6. The communication system of claim 5.

[0185] (Appendix 7) A computer provided in a communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A control program that executes processing.

[0186] (Appendix 8) Identifying the standby node that satisfies a cell number requirement in addition to the latency requirement from the standby node group based on a number of cells that the accelerator can accommodate in addition to the latency and the location information; 8. The control program of claim 7, which causes the computer to execute a process.

[0187] (Appendix 9) As the location information, the standby node is identified from a first standby node group provided in the same first server rack as the failed node. 9. The control program according to claim 7 or 8, which causes the computer to execute a process.

[0188] (Appendix 10) When the first group of standby nodes does not exist, the standby node is identified from a second group of standby nodes that is provided in a second server rack different from the first server rack and whose latency with the accelerator in the first server rack satisfies the latency requirement. 10. The control program of claim 9, which causes the computer to execute a process.

[0189] (Appendix 11) When the second group of standby nodes does not exist, identifying the standby node from a third group of standby nodes that is provided in the second server rack and whose latency with the accelerator in the second server rack satisfies the latency requirement. 11. The control program of claim 10, which causes the computer to execute a process.

[0190] (Appendix 12) When the third group of standby nodes does not exist, the standby node is identified from a fourth group of standby nodes that is provided in the second server rack and whose latency with the accelerator in a third server rack different from the first server rack or the second server rack satisfies the latency requirement. 12. The control program of claim 11, which causes the computer to execute a process.

[0191] (Appendix 13) A computer provided in a communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A control program that executes processing.

[0192] (Appendix 14) Identifying the standby node that satisfies a cell number requirement in addition to the latency requirement from the standby node group based on a number of cells that the accelerator can accommodate in addition to the latency and the location information; The control program according to claim 13, wherein the processing is executed by the computer.

[0193] (Appendix 15) As the location information, the standby node is identified from a first standby node group provided in the same first server rack as the failed node. 15. The control program according to claim 13 or 14, wherein the processing is executed by the computer.

[0194] (Appendix 16) When the first group of standby nodes does not exist, the standby node is identified from a second group of standby nodes that is provided in a second server rack different from the first server rack and whose latency with the accelerator in the first server rack satisfies the latency requirement. 16. The control program according to claim 15, wherein the processing is executed by the computer.

[0195] (Appendix 17) When the second group of standby nodes does not exist, identifying the standby node from a third group of standby nodes that is provided in the second server rack and whose latency with the accelerator in the second server rack satisfies the latency requirement. 17. The control program according to claim 16, wherein the processing is executed by the computer.

[0196] (Appendix 18) When the third group of standby nodes does not exist, the standby node is identified from a fourth group of standby nodes that is provided in the second server rack and whose latency with the accelerator in a third server rack different from the first server rack or the second server rack satisfies the latency requirement. 18. The control program according to claim 17, wherein the processing is executed by the computer. [Explanation of symbols]

[0197] 100,200: Virtualized base station system 1: Resource Manager 11,211:CPU 111: Latency table creation section 112: Accelerator table creation section 113: Base station requirement value setting unit 12: Memory 13: Display control device 131:Display device 14:Storage device 141: Resource table 142: Latency table 143: Accelerator table 144:Base station requirements 145: Latency requirement value 146: Cell count requirement value 114: Accelerator performance setting unit 147: Cell performance value 15: Input IF 151: Mouse 152: Keyboard 16: External recording medium processing device 160: Recording media 17: Communication Interface 2: Server rack 20: Base station server 21 : Node 22: Fabric SW 23: Component Box 231: Accelerator 232: GPU 201: Latency measurement unit 202:Base station function notification section 3: Network switch 30: MEC server

Claims

1. A communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A communication system comprising a control unit.

2. The control unit is Identifying the standby node that satisfies a cell number requirement in addition to the latency requirement from the standby node group based on a number of cells that the accelerator can accommodate in addition to the latency and the location information; The communication system according to claim 1 .

3. The control unit is As the location information, the standby node is identified from a first standby node group provided in the same first server rack as the failed node.

3. A communication system according to claim 1 or 2.

4. The control unit is When the first group of standby nodes does not exist, the standby node is identified from a second group of standby nodes that is provided in a second server rack different from the first server rack and whose latency with the accelerator in the first server rack satisfies the latency requirement. The communication system according to claim 3.

5. The control unit is When the second group of standby nodes does not exist, the method identifies the standby node from a third group of standby nodes that is provided in the second server rack and whose latency with the accelerator in the second server rack satisfies the latency requirement.

5. The communication system according to claim 4.

6. The control unit is When the third standby node group does not exist, the standby node is identified from a fourth standby node group that is provided in the second server rack and in which the latency with the accelerator in a third server rack different from the first server rack or the second server rack satisfies the latency requirement.

6. The communication system according to claim 5.

7. A computer provided in a communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A control program that executes processing.

8. A computer provided in a communication system including a base station having a Composable Disaggregated Infrastructure (CDI) configuration, storing measurement results of latencies between all nodes and all accelerators included in the communication system; When a node failure occurs, a standby node that satisfies a latency requirement is identified from a group of standby nodes based on the latency and location information of an accelerator connected to the failed node; Switching the failed node to the standby node; A control method for performing processing.

Images