Bare metal server and bare metal server cluster

Abstract

The invention provides a bare metal server and a bare metal server cluster. The bare metal server comprises at least two computing daughter cards, a gating backboard and a processor mainboard, wherein the gating backboard is respectively connected with the at least two computing daughter cards and the processor mainboard, and the gating backboard is used for distributing a target computing daughter card in the at least two computing daughter cards to an application terminal according to a terminal application instruction. According to the invention, a new mechanism of the bare metal server is provided, so that the computing power is pooled, computing power resources can be flexibly configured, and the situation of resource waste or insufficiency is reduced. The server dynamically obtains or releases the resources from the computing power pool according to the actual computing power load, so that the utilization rate of the computing power resources is maximized, data physical isolation between cloud users can be realized, and the data security is improved.

Description

Technical Field

[0001] The present application relates to the field of server technology, and in particular to a bare metal server and a bare metal server cluster. Background Art

[0002] Cloud computing has gradually become a mainstream application because it provides remote computing resource services. In the early days, the focus of cloud computing was on the layout of server hardware to meet user needs. With the development of cloud computing, the cost of servers, that is, the number of deployments and energy efficiency issues have gradually emerged. In addition, the current cloud computing operation model is to provide computing resources to cloud customers. Most of these resources are shared hardware resources, shared in the form of software virtual machines, which will also cause performance loss. In addition, some scenarios with high data security requirements, such as finance and banking, must use hardware isolation, and there are problems with using virtual machine services.

[0003] However, the current server architecture that provides computing power for cloud computing is implemented through a powerful CPU (Central Processing Unit / Processor) configured with several smart cards, which are directly connected to each other, and resource allocation to multiple cloud users is implemented through software virtual machines. This technology can easily threaten the security of the server, and computing power resources are not allocated in the best way. Summary of the invention

[0004] In order to solve the above technical problems, the present application proposes a bare metal server and a bare metal server cluster.

[0005] To solve the above technical problems, the present application proposes a bare metal server, which includes: at least two computing daughter cards, a gate backplane, and a processor mainboard; The gating backplane is respectively connected to the at least two computing daughter cards and the processor mainboard, and the gating backplane is used to allocate a target computing daughter card among the at least two computing daughter cards to the application terminal according to the terminal application instruction.

[0006] The processor mainboard calculates the number of computing subcards required for the terminal to apply for the instruction according to the computing power specifications of the at least two computing subcards; The gating backplane determines the target computing daughter card according to the number of the computing daughter cards.

[0007] The gating backplane acquires the remaining computing power of each computing sub-card of the at least two computing sub-cards, and selects the target computing sub-card according to the remaining computing power of the at least two computing sub-cards.

[0008] Wherein, the bare metal server includes at least two processor motherboards; The gated backplane user allocates a target processor motherboard among the at least two processor motherboards to the target computing daughter card according to the computing power environment of the target computing daughter card.

[0009] The gating backplane acquires the remaining processing resources of each of the at least two processor mainboards, and selects a target processor main card according to the remaining processing resources of the at least two processor mainboards.

[0010] The computing daughter card and / or the processor mainboard are installed on the bare metal server in a pull-out manner.

[0011] Wherein, the strobe backplane includes a high-speed chip, and the high-speed chip is respectively connected to the processor mainboard and the computing daughter card through an I / O bus.

[0012] In order to solve the above technical problems, the present application also proposes a bare metal server cluster, wherein the bare metal server cluster includes at least two bare metal servers; the at least two bare metal servers are connected via a gated backplane; The bare metal server is the above-mentioned bare metal server.

[0013] When the computing power pool of one of the bare metal servers is insufficient, a computing power application request is made to other bare metal servers through the selection backplane, so as to realize the service of the application terminal through the computing sub-cards of the other bare metal servers.

[0014] When the processor mainboard of one of the bare metal servers is abnormal, the idle processor mainboards of its own bare metal server and / or other bare metal servers are obtained through the selection backplane to migrate the mainboard resources of the abnormal processor mainboard to the idle processor mainboard.

[0015] Compared with the prior art, the beneficial effects of the present application are as follows: the present application proposes a bare metal server comprising: at least two computing sub-cards, a gating backplane, and a processor mainboard; wherein the gating backplane is respectively connected to the at least two computing sub-cards and the processor mainboard, and the gating backplane is used to allocate the target computing sub-card of the at least two computing sub-cards to the application terminal according to the terminal application instruction. The present application proposes a new bare metal server mechanism to pool computing power, flexibly configure computing power resources, and reduce resource waste or shortage. The server can dynamically obtain or release resources from the computing power pool according to the actual computing power load to maximize the utilization of computing power resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 It is a schematic diagram of the framework of an embodiment of a bare metal server provided by the present application; Figure 2 It is a schematic diagram of the framework of an embodiment of a bare metal server cluster provided by the present application; Figure 3 This is a cluster topology diagram of the bare metal server cluster provided in this application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can, for example, be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] In order to solve the above problems, this application is based on the concept of bare metal servers, and makes computing power pooling. Multiple CPUs share the computing power pool. Under the condition of physical data isolation, the server can dynamically obtain or release resources from the computing power pool according to the actual computing power load, so as to maximize the utilization of computing power resources. At the same time, a cluster method for multiple servers to share the computing power pool is proposed.

[0020] Specifically, the bare metal server involved in this application, also known as a single-tenant environment physical server, refers to one or more independent servers that provide exclusive physical computing resources. It is an innovative product in the server field, combining the elasticity of virtual machines and the performance advantages of physical machines to become a dedicated physical server in the cloud. Bare metal servers directly provide computing resources at the hardware level without going through the virtualization layer, thereby avoiding the performance loss caused by virtualization and providing users with ultimate computing performance and resource exclusivity.

[0021] This application is improved on the basis of the existing bare metal server. The server architecture of this application is realized by several small CPUs and several smart cards, which are interconnected through the I / O enabling backplane. The I / O enabling backplane allows the CPU to connect to different numbers of smart cards. The resource allocation for multiple cloud users is directly implemented on different CPUs without going through virtual machines, thus realizing hardware isolation. Among them, the I / O bus refers to the cable and connector system used to transmit data and control signals specified by the I / O path technology. ‌

[0022] Please refer to Figure 1 , Figure 1 It is a schematic diagram of the framework of an embodiment of a bare metal server provided in the present application.

[0023] like Figure 1 As shown, the bare metal server 100 provided in the present application includes: at least two computing daughter cards 11, a gate backplane 12, and a processor mainboard 13, that is, Figure 1 The CPU motherboard is shown.

[0024] Among them, the I / O enabling backplane 12 of the present application is respectively connected to at least two computing sub-cards 11 and the CPU mainboard 13, and the function of the I / O enabling backplane 12 is mainly used to allocate the target computing sub-card among at least two computing sub-cards 11 to the application terminal according to the terminal application instruction.

[0025] Specifically, the application terminal includes but is not limited to: a smart phone, a personal computer, a tablet computer, and a vehicle-mounted system. The application terminal submits a terminal application instruction to the bare metal server 100 via Ethernet to apply for computing resources of the bare metal server 100.

[0026] The CPU mainboard 13 in the bare metal server 100 transmits the terminal application instruction to the I / O enabling backplane 12. The I / O enabling backplane 12 selects one or more target computing subcards 11 from a number of computing subcards 11 according to the terminal application instruction as operator resources allocated to the applying terminal.

[0027] The CPU mainboard 13 provides an environment for the application terminal to run the operator resources of the target computing subcard 11 by constructing an operating system.

[0028] Specifically, the I / O gating backplane 12 selects one or more target computing daughter cards 11 from a plurality of computing daughter cards 11 according to the terminal application instruction in the following manner: The I / O enabling backplane 12 obtains the remaining computing power of all connected computing sub-cards 11, and then determines the required amount of operator resources applied for by the applying terminal according to the terminal application instruction. Finally, the I / O enabling backplane 12 determines the target computing sub-card 11 according to the required amount of operator resources applied for by the applying terminal and the remaining computing power of all computing sub-cards 11, so that the total amount of operator resources of the target computing sub-card 11 meets the required amount of operator resources applied for by the applying terminal.

[0029] The I / O enabling backplane 12 preferentially selects a computing daughter card 11 whose remaining computing power is less than the original computing power as a target computing daughter card 11 to improve the utilization rate of the computing daughter card 11 .

[0030] Specifically, the I / O gating backplane 12 may select one or more target computing daughter cards 11 from a plurality of computing daughter cards 11 according to the terminal application instruction in the following manner: The CPU mainboard 13 calculates the number of computing subcards required for the terminal application instruction according to the computing power specifications of all two computing subcards 11. Then, the I / O selection backplane 12 determines the target computing subcard 11 according to the number of computing subcards calculated by the CPU mainboard 13.

[0031] like Figure 1 As shown, the bare metal server 100 of the present application includes at least two CPU mainboards 13.

[0032] There are several high-speed connectors on the I / O enabling backplane 12, which are respectively interconnected with several CPU mainboards 13 and several computing sub-cards 11. The I / O enabling backplane 12 serves as a bridge between the CPU mainboard 13 and the computing sub-card 11, and is designed with a high-speed chip (such as FPGA, Field Programmable Gate Array) for interconnection. The CPU mainboard 13 and the high-speed chip (such as FPGA) are interconnected through an I / O bus (such as PCIE, peripheral component interconnect express, USB, Universal SerialBus, etc.), and the computing sub-card 11 and the high-speed chip (such as FPGA) are also interconnected through an I / O bus (such as PCIE, USB, etc.). Each CPU mainboard 13 is connected to the Ethernet through a network bus (such as an electrical port, an optical port, etc.).

[0033] Each CPU motherboard 13 is an independent hardware resource, isolated from each other and not connected to each other. They are the main servers provided by the cloud server to cloud users. All computing daughter cards 11 form a unified computing power pool, which is uniformly controlled and managed by the high-speed chip (such as FPGA) on the I / O selection backplane 12.

[0034] After purchasing CPU resources, cloud users apply for computing resources at the terminal. The CPU mainboard 13 applies to the I / O enabling backplane 12 for the required computing sub-cards according to the computing power specifications. The high-speed chip (such as FPGA) on the I / O enabling backplane 12 configures the I / O bus and enables the number of computing sub-cards corresponding to the computing resources required by the cloud user.

[0035] The bare metal server 100 of the present application can realize the sharing of the computing resource pool of a bare metal server, and the servers are physically isolated from each other, thereby avoiding resource sharing and resource waste of virtual machines.

[0036] In particular, the CPU specification of the CPU mainboard 13 can be reduced, and the main computing power is provided by the computing daughter card 11, thereby reducing the cost while keeping the computing power unchanged.

[0037] Among them, in the prior art, servers and computing power pooling are basically software virtualization. This application is based on hardware virtualization, which reduces software overhead and improves software efficiency compared to software virtualization.

[0038] Furthermore, the bare metal server 100 of the present application can also select one or more CPU motherboards 13 from multiple CPU motherboards 13 as the target CPU motherboard 13 to provide a running environment for the operator resources applied for by the application terminal.

[0039] Specifically, the I / O gating backplane 12 selects one or more target CPU mainboards 13 from a plurality of CPU mainboards 13 according to the terminal application instruction in the following manner: The I / O enabling backplane 12 obtains the remaining processing resources of all connected CPU mainboards 13, and then determines the required processing resources requested by the applying terminal according to the terminal application instruction. Finally, the I / O enabling backplane 12 determines the target CPU mainboard 13 according to the required processing resources requested by the applying terminal and the remaining computing power of all CPU mainboards 13, so that the total processing resources of the target CPU mainboard 13 meet the required processing resources requested by the applying terminal.

[0040] The I / O enabling backplane 12 preferentially selects a CPU mainboard 13 whose remaining processing resources are less than the original processing resources as the target CPU mainboard 13 to improve the utilization rate of the CPU mainboard 13 .

[0041] Furthermore, the CPU motherboard 13 and the computing daughter card 11 are designed to be removable, which is convenient for maintenance personnel to replace. The CPU motherboard 13 monitors abnormalities through the BMC, and the computing daughter card 11 monitors through the high-speed chip of the I / O gate backplane 12.

[0042] Among them, the BMC (Baseboard Management Controller) is a controller on the server motherboard, which is responsible for remote access, power management and system control of the server. If the BMC monitoring is abnormal, it may make it impossible to remotely manage the server or collect important status information of the server.

[0043] exist Figure 1 Based on the bare metal server shown in the figure, in order to prevent all CPU motherboards on the bare metal server from requiring a large amount of computing resources at the same time, resulting in the exhaustion of the computing power pool, this application also provides a bare metal server cluster. For details, please refer to Figure 2 , Figure 2 It is a framework diagram of an embodiment of a bare metal server cluster provided by the present application.

[0044] like Figure 2 As shown, the bare metal server cluster 200 provided in the present application includes at least two bare metal servers 21 , wherein the at least two bare metal servers 21 are connected via a gate backplane 212 .

[0045] Specifically, the specific framework of the bare metal server 21 of this application can be found in Figure 1 The bare metal server 100 shown is not described in detail here.

[0046] The I / O gate backplane 212 of each bare metal server 21 can form a cluster through the Ethernet protocol or the IB (InfiniBand, wireless bandwidth) protocol, that is, the high-speed chips (such as FPGA) are interconnected. When the computing power pool of one of the bare metal servers 21 is exhausted, the high-speed chip (such as FPGA) submits a computing power application request to other bare metal servers 21 to prevent shortage.

[0047] In particular, multiple bare metal servers 21 can use a full mesh (all nodes are directly connected) topology to achieve clustering, allowing a bare metal server 21 with exhausted computing power pool to request computing power from multiple other bare metal servers 21 at the same time. The topology is as follows: Figure 3 The diagram shows a cluster connection mode of four bare metal servers 21.

[0048] Specifically, when the BMC on the CPU motherboard 211 finds that there is an abnormality in the CPU motherboard 211, it will issue an abnormality alarm to the high-speed chip of the I / O enabling backplane 212. The high-speed chip will search for the idle CPU motherboards 211 of this bare metal server 21 and other bare metal servers 21 in the cluster, and migrate the resources of the abnormal CPU motherboard 211 to the new idle CPU motherboard 211, and at the same time alarm the operation and maintenance personnel to replace the CPU motherboard 211.

[0049] Specifically, when the high-speed chip monitoring of the I / O enabling backplane 212 finds that there is an abnormality in the computing sub-card 213, such as insufficient operators, an abnormality alarm will be issued to the I / O enabling backplane 212, and the high-speed chip will search for idle computing sub-cards 213 of the bare metal server 21 and other bare metal servers 21 in the cluster, and allocate the idle computing sub-cards 213 to the applying terminal to realize the services of the applying terminal.

[0050] The bare metal server cluster 200 of the present application does not need to coordinate operators and processing resources through the CPU motherboard 211 of other bare metal servers 21, but can be directly implemented from the I / O enabling backplane 212, which will not occupy the resources of other CPU motherboards 211 and improve efficiency.

[0051] This application proposes a new bare metal cloud computing server structure, which pools computing power and allows multiple CPUs to share the computing power pool. Computing power resources can be flexibly configured to reduce resource waste or shortage. The server can dynamically obtain or release resources from the computing power pool according to the actual computing power load, maximizing the utilization of computing power resources.

[0052] This application eliminates the concept of virtual machines. Each cloud user uses independent CPU resources, which have weak computing power. The main computing power is provided by the computing power pool. It can achieve physical isolation of data between cloud users and improve data security, especially in the financial, banking and other industries.

[0053] This application proposes a cluster solution for sharing computing power. When the computing power pool of a server is exhausted, computing power resources are requested from other servers. All servers are interconnected through a full mesh topology, allowing a server with an exhausted computing power pool to request computing power from multiple other servers at the same time.

[0054] Since this application uses a computing power pool, the CPU can use a weaker chip, which reduces the cost of the server. Through the distributability of computing power, the cost of total computing power resources is reduced.

[0055] The present application proposes a backup method after a CPU anomaly, which reduces the maintenance cost generated during the anomaly.

[0056] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A bare metal server, characterized in that: The bare metal server includes: at least two computing daughter cards, a gate backplane, and a processor mainboard; The gating backplane is respectively connected to the at least two computing daughter cards and the processor mainboard, and the gating backplane is used to allocate a target computing daughter card among the at least two computing daughter cards to the application terminal according to the terminal application instruction.

2. The bare metal server according to claim 1, characterized in that: The processor mainboard calculates the number of computing subcards required for the terminal to apply for the instruction according to the computing power specifications of the at least two computing subcards; The gating backplane determines the target computing daughter card according to the number of the computing daughter cards.

3. The bare metal server according to claim 1 or 2, characterized in that: The gating backplane acquires the remaining computing power of each computing sub-card of the at least two computing sub-cards, and selects a target computing sub-card according to the remaining computing power of the at least two computing sub-cards.

4. The bare metal server according to claim 1, characterized in that: The bare metal server includes at least two processor motherboards; The gated backplane user allocates a target processor motherboard among the at least two processor motherboards to the target computing daughter card according to the computing power environment of the target computing daughter card.

5. The bare metal server according to claim 4, characterized in that: The gating backplane acquires the remaining processing resources of each of the at least two processor mainboards, and selects a target processor main card according to the remaining processing resources of the at least two processor mainboards.

6. The bare metal server according to claim 1, characterized in that: The computing daughter card and / or the processor mainboard are installed on the bare metal server in a pull-out manner.

7. The bare metal server according to claim 1, characterized in that: The gate backplane includes a high-speed chip, and the high-speed chip is respectively connected to the processor mainboard and the computing daughter card through an I / O bus.

8. A bare metal server cluster, characterized in that: The bare metal server group includes at least two bare metal servers; the at least two bare metal servers are connected via a gate backplane; Wherein, the bare metal server is the bare metal server described in any one of claims 1 to 7.

9. The bare metal server cluster according to claim 8, characterized in that: When the computing power pool of one of the bare metal servers is insufficient, a computing power application request is made to other bare metal servers through the selection backplane, so as to realize the service of the application terminal through the computing sub-cards of the other bare metal servers.

10. The bare metal server cluster according to claim 8, characterized in that: When the processor motherboard of one of the bare metal servers is abnormal, the idle processor motherboards of its own bare metal server and / or other bare metal servers are obtained through the selection backplane to migrate the motherboard resources of the abnormal processor motherboard to the idle processor motherboard.

Images