CN102110071B - Virtual machine cluster system and implementation method thereof - Google Patents

Abstract

The invention discloses a virtual machine cluster system and a realization method thereof, so as to overcome the defect of high cost of the current cluster system. The virtual machine cluster system is built on multiple physical servers, running virtual machines on these physical servers to provide multiple virtual computing nodes, these virtual computing nodes are divided into a management node and multiple computing nodes, wherein: the management node, used to create a management library to record the host name and dynamic IP address of each computing node, and manage the entire virtual machine cluster system according to these host names and dynamic IP addresses; these computing nodes are used to complete the The computing task of the virtual machine cluster system. The technical scheme of the present invention uses a virtual machine to build a cluster system, which reduces the number of physical servers used and reduces the cost of the cluster system.

Description

A virtual machine cluster system and its implementation method

technical field

The invention relates to the field of computer applications, in particular to a virtual machine cluster system and an implementation method thereof.

Background technique

A computer cluster system is a computing system that is connected by a group of loosely integrated computer software or hardware and highly closely cooperates to complete computing work. The high-availability cluster of the cluster system generally means that when a node in the cluster system fails, the tasks running on it will be automatically migrated to other normal running nodes. This migration process does not affect the operation of the entire cluster system.

The hardware composition of a traditional cluster system generally includes server groups and shared storage devices. A server group generally includes multiple server nodes, and operating systems and cluster software programs need to be installed uniformly; shared storage devices are used to ensure data consistency among server nodes. In addition, in order to ensure the high availability of the cluster system, a redundant backup method is generally adopted, so that one server needs a backup device, or prepare multiple backup devices for each server (called one more backup). These backup devices are also one of the reasons for the high cost of the entire cluster system.

With the development of science and technology, the demand for small-scale cluster systems is also gradually increasing, and cluster systems are required to complete high-performance computing, but the small scale also determines the need to minimize the economic cost of the entire cluster system. Therefore, the high cost of the cluster system has become one of the obstacles affecting the development of the cluster system.

Contents of the invention

The technical problem to be solved by the present invention is to provide a virtual machine cluster system to overcome the defect of high cost of the current cluster system.

In order to solve the above technical problems, the present invention provides a virtual machine cluster system, the virtual machine cluster system is constructed on multiple physical servers, running virtual machines on these physical servers to provide multiple virtual computing nodes, these virtual Computing nodes are divided into a management node and multiple computing nodes, among which:

The management node is used to create a management library to record the hostname and dynamic IP address of each computing node, and manage the entire virtual machine cluster system according to the hostname and dynamic IP address;

The computing nodes are used to complete computing tasks of the virtual machine cluster system.

Preferably, these physical servers are used to provide disk images of virtual machines in the copy-on-write format of Qemu.

Preferably, the management node includes:

A Dynamic Host Configuration Protocol server for dynamically distributing the dynamic IP address;

Domain name server, used to complete domain name resolution;

The network file system server is used to share the disk image.

Preferably, the management node includes:

The extended server is configured to configure a physical server newly connected to the virtual machine cluster system as a physical server running computing nodes.

Preferably, the physical servers are used to share disks using a network file system.

Preferably, heartbeat connections are established between the management node and the computing nodes, and the management node periodically sends heartbeat signals to the computing nodes, and performs virtual machine migration operations on computing nodes that have not received heartbeat feedback.

In order to solve the above-mentioned technical problems, the present invention also provides a method for realizing a virtual machine cluster system, the virtual machine cluster system is built on multiple physical servers, running virtual machines on these physical servers to provide multiple virtual computing nodes , these virtual computing nodes are divided into a management node and multiple computing nodes, among which:

Create a management library on the management node to record the host name and dynamic IP address of each computing node, and manage the entire virtual machine cluster system according to these host names and dynamic IP addresses;

Wherein, the computing nodes are used to complete computing tasks of the virtual machine cluster system.

Preferably, these physical servers provide disk images of virtual machines in the copy-on-write format of Qemu.

Preferably, the physical servers share disks using a network file system.

Preferably, a heartbeat connection is established between the management node and the computing nodes.

Compared with the prior art, the technical scheme of the present invention adopts a virtual machine to build a cluster system, which reduces the number of physical servers used and reduces the cost of the cluster system. The technical scheme of the present invention can monitor the health state of the virtual machine through the heartbeat connection, start the dynamic migration of the virtual machine in good time, ensure the uninterrupted operation of the service on the virtual machine, and achieve high availability of the entire virtual machine cluster system. The technical solution of the present invention completes the management and configuration of the cluster system through the management nodes, and the computing nodes are responsible for computing tasks. Such an architecture can dynamically add computing nodes without restarting the system, and the entire virtual machine cluster system has high scalability. In addition, in the technical solution of the present invention, all virtual machines use shared disk mirroring, which greatly reduces the storage space of computing nodes, and uses NFS to share data storage, without using shared storage devices alone, further reducing the entire virtual machine. The cost of the machine cluster system.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

Description of drawings

The accompanying drawings are used to provide a further understanding of the technical solution of the present invention, and constitute a part of the description, and are used together with the embodiments of the present invention to explain the technical solution of the present invention, and do not constitute a limitation to the technical solution of the present invention. In the attached picture:

1 is a schematic diagram of the composition of a virtual machine cluster system according to an embodiment of the present invention;

FIG. 2 is a system view of virtualization using KVM in Embodiment 1 of the present invention.

Detailed ways

The implementation of the present invention will be described in detail below in conjunction with the accompanying drawings and examples, so as to fully understand and implement the process of how to apply technical means to solve technical problems and achieve technical effects in the present invention.

First of all, if there is no conflict, the embodiment of the present invention and the combination of various features in the embodiment are within the protection scope of the present invention. In addition, the steps shown in the flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and, although a logical order is shown in the flow diagrams, in some cases, the sequence may be different. The steps shown or described are performed in the order herein.

With the development of virtualization technology, it has become a reality to run multiple operating systems on one server at the same time. Hardware virtualization technology makes the performance of the virtual machine reach a level close to that of the host. Using virtualization technology, the cluster system that originally required dozens or even more servers to build can now be realized with only half or even fewer servers, and the overall development of virtualization technology is also to ensure the high availability and high availability of the entire cluster system. High scalability provides technical assurance.

Embodiment 1, virtual machine cluster system

FIG. 1 is a schematic diagram of the composition of the first embodiment. As shown in Figure 1, the virtual machine cluster system of this embodiment is built on multiple physical servers 120 (which can be accessed through the network on ordinary physical machines 110), and the present invention refers to these physical servers as virtualization services Provider (Virtualization Service Provider, VSP), running virtual machines on these physical servers to provide multiple virtual computing nodes (Virtual Computer Node, VCN) 130, these virtual computing nodes are divided into management nodes 140 and computing nodes 150 In this case, the number of management node 140 is one, and these virtual computing nodes are all computing nodes 150 except the management node 140, wherein:

There is only one management node 140 in the entire virtual machine cluster system, which is used to configure and manage the entire virtual machine cluster system; a management library is created on it to record the host name and dynamic IP address of each computing node, according to The host name and dynamic IP address of each computing node recorded in the management library can realize the management of the entire virtual machine cluster system;

There are multiple computing nodes 150 in the entire virtual machine cluster system, connected to the management node 140, and used to complete computing tasks of the virtual machine cluster system.

In this embodiment, the structure of the physical server (i.e. VSP) 120 is shown in Figure 2, which adopts hardware virtualization technology (Kernel based Virtual Machine, KVM), and creates a special character device /dev in the system by loading the kernel. /kvm, to communicate with userland device models. Through this device, the address space of the guest operating system is independent of the address space of the kernel or other running guest operating systems. And by introducing a client mode, the client has its own private address space, which is mapped when the guest operating system (Guest OS) is instantiated, and the client address is mapped on this private address space. After the KVM is loaded, the guest operating system can be started in the user space, that is, the provision of the virtualization service is completed. It should be noted that the creation of a special character device /dev/kvm in the system can be realized by using the existing technology, which is not specifically limited in the present invention.

Figure 2 is a system view of virtualization using KVM. The hardware platform (hardware) at the bottom is a hardware platform that can support hardware virtualization. On the hardware platform is a Linux operating system (OS), which contains a virtual machine. The monitoring program hypervisor. On such a platform, the client operating system (that is, Guest OS) can be run, and the same application programs as ordinary operating systems can be run in the Guest OS. It should be noted that the technical content of using KVM for virtualization shown in FIG. 2 is a brief introduction to the prior art, which is not described in detail in the present invention.

The physical server 120 is used to provide the disk image of the virtual machine in Qemu's copy-on-write (Qemu Copy-On-Write, QCOW) format, and the QCOW format supports a snapshot mode. In this snapshot mode, all write operations of the virtual machine to the disk image will be written separately to an additional temporary file instead of being written to the disk image. This sharing mode supports all virtual machines in the entire virtual machine cluster system to use the same disk image without data inconsistency, which greatly reduces the storage space (Storage) requirements of the entire virtual machine cluster system , reducing the cost of storage devices.

In addition, the cluster systems in the prior art realize that all physical servers use a unified storage space by additionally using a shared storage device. In the virtual machine cluster system of the present invention, the physical server 120 is used to use NFS to share disks. NFS is a distributed file system that allows a local computer to use the devices of computers on the network as if they were local devices, and only needs to mount the computer devices on the network to the local system. In this way, only a disk image needs to be placed on one VSP, and other VSPs do not need a large disk space, which not only reduces the shared storage device, but also reduces the size of the required storage space, and achieves the purpose of reducing the cost of the cluster system.

In the technical solution of the present invention, a heartbeat connection can be established between the management node 140 and the computing node 150, and the management node 140 periodically sends a heartbeat signal to all computing nodes 150, and receives the corresponding heartbeat feedback from the computing node 150, indicating that the computing node 150 active or normal. If the management node 140 does not receive the heartbeat feedback from the computing node 150, it can be considered that the corresponding computing node 150 is in an abnormal state (such as a fault state or a dead state, etc.), and such a state may be due to a failure of the VSP or its own Due to the fault, the virtual machine migration operation is started at this time, and the computing node is migrated to the VSP in a normal state to continue running. This ensures the highly reliable operation of the entire system. In this way, the high availability of the entire virtual machine cluster system is well guaranteed through the dynamic migration of virtual machines.

The management node 140 is mainly used to manage the entire virtual machine cluster system and the computing node 150 is mainly used to complete the division of computing tasks. Specifically, for example, if the Condor (an open source tool) system is deployed in this embodiment, then the management Configure condor_collector, condor_negotiator, and condor_schedd on node 140, and configure condor_startd on computing node 150.

Since all computing nodes share a disk image, static IP cannot be used, because static IP will write the IP address into the disk image. Therefore, the management node 140 includes a Dynamic Host Configuration Protocol (Dynamic Host Configuration Protocol, DHCP) server, a Domain Name Server (Domain Name Server, DNS) and a Network File System (Network File System, NFS) server, wherein:

The DHCP server is used for dynamically distributing IP addresses;

The DNS server is used to complete domain name resolution;

The NFS server is used to share the disk images to each VSP, and these disk images are set to a read-only mode, and each computing node uses a snapshot mode to access the disk images.

The management node 140 may also include an extension server, where the extension server is used to configure a physical server newly connected to the virtual machine cluster system as a VSP running a computing node. This management mode can easily expand the scale of the virtual machine cluster system. It only needs to configure the newly added physical server as a VSP running the computing node, and then perform appropriate configuration. The operation of adding computing nodes on the management node 140 is added to the entire cluster computing system. During this process, the dynamic addition of newly added physical servers can be realized without restarting the virtual machine cluster system or the management node 140 .

Embodiment 2, the realization method of the virtual machine cluster system

In combination with the virtual machine cluster system shown in FIG. 1 and its description, this embodiment implements the virtual machine cluster system, which is built on multiple physical servers, and runs virtual machines on these physical servers to provide multiple virtual computing nodes. These virtual computing nodes are divided into a management node and multiple computing nodes, among which:

Create a management library on the management node to record the host name and dynamic IP address of each computing node, and manage the entire virtual machine cluster system according to these host names and dynamic IP addresses;

Wherein, the computing nodes are used to complete computing tasks of the virtual machine cluster system.

Wherein, these physical servers provide the disk image of the virtual machine in the copy-on-write format of Qemu.

Wherein, these physical servers use a network file system to share disks.

Wherein, a heartbeat connection is established between the management node and the computing nodes.

For the implementation method of the virtual machine cluster system provided in this embodiment, please refer to the technical solution of the virtual machine cluster system provided in the first embodiment for understanding.

Traditional cluster systems are established through physical servers and shared storage, and these devices require high-cost investment. The technical solution of the present invention uses a virtual machine to build a cluster system, which not only reduces the number of physical servers and improves the utilization rate of the physical server, but also can use virtual machine technology to improve the availability and scalability of the cluster system and significantly reduce the cost. At the same time, the purpose of high-performance computing of the cluster system can also be achieved.

In the technical solution of the present invention, the dynamic migration technology of the virtual machine is used to realize the high availability of the cluster system. Virtual machine live migration means that during the process of virtual machine migration, services running in the virtual machine will not be interrupted. Virtual machine dynamic migration is to use iterative copy method to transfer virtual machine memory status, and dynamically migrate virtual machines (or computing nodes) that have failed or are in other abnormal working states to other VSPs, ensuring the continuity of services and realizing High availability of clustered systems.

The technical scheme of the invention adopts a composition mode of a management node and multiple computing nodes, which facilitates the dynamic expansion of the cluster system. If you need to expand the scale of the cluster system, you only need to add the corresponding physical server, build a computing node on it, and then maintain the configuration information on the management node to expand the newly added physical server to the existing cluster system. Without restarting the cluster system or management nodes, the computing power of the cluster system can be rapidly improved, and the computing performance increases almost linearly with the increase in the number of physical servers.

The technical scheme of the invention makes it unnecessary to prepare additional backup equipment for the high availability of the cluster system proposed by the invention, thereby saving the cost of the backup equipment. In addition, all the virtual machines of the cluster system proposed by the present invention adopt a shared image mode, which greatly reduces the demand and use of storage space. In addition, the cluster system provided by the present invention shares the disk space through NFS, and does not need to separately purchase a shared storage device, thereby reducing the cost of the storage space.

Those skilled in the art should understand that each component of the above-mentioned system embodiment of the present invention or each step of the method embodiment can be implemented by a general-purpose computing device, and they can be concentrated on a single computing device, or distributed in multiple On a network composed of computing devices, optionally, they can be implemented with executable program codes of computing devices, thus, they can be stored in storage devices and executed by computing devices, or they can be made into individual integrated circuit modules, or by making multiple modules or steps among them into a single integrated circuit module. As such, the present invention is not limited to any specific combination of hardware and software.

Although the embodiments disclosed in the present invention are as above, the described content is only an embodiment adopted for the convenience of understanding the present invention, and is not intended to limit the present invention. Anyone skilled in the technical field to which the present invention belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed by the present invention, but the patent protection scope of the present invention, The scope defined by the appended claims must still prevail.

Claims (9)

1. virtual machine cluster system, it is characterized in that, this virtual machine cluster system is structured on a plurality of physical servers, move virtual machines so that a plurality of virtual computing nodes to be provided at those physical servers, those virtual computing nodes are divided into a management node and a plurality of computing node, virtual computing nodes all in the whole virtual machine cluster system all use same disk mirroring, wherein:

This management node is used for creating host name and the dynamic IP addressing of a management holder to record each computing node, according to those host name and dynamic IP addressing, whole virtual machine cluster system is managed; This management node comprises expansion servers, and this expansion servers is configured to move the virtualization services supplier VSP of computing node for the physical server that will newly access this virtual machine cluster system;

Those computing nodes are for the calculation task of finishing this virtual machine cluster system.

2. virtual machine cluster system according to claim 1 is characterized in that:

Those physical servers provide the disk mirroring of virtual machine for the Copy on write form that adopts Qemu.

3. virtual machine cluster system according to claim 2 is characterized in that, this management node also comprises:

Dynamic Host Configuration Protocol server is for the dynamic distribution that carries out this dynamic IP addressing;

Name server is used for finishing domain name mapping;

Network file system server is used for this disk mirroring is shared.

4. virtual machine cluster system according to claim 1 is characterized in that:

Those physical servers are used for using the network file system(NFS) shared disk.

5. virtual machine cluster system according to claim 1 is characterized in that:

Setting up between this management node and those computing nodes has heartbeat to be connected, and this management node regularly sends heartbeat signal to those computing nodes, and the computing node of not receiving the heartbeat feedback is carried out virtual machine (vm) migration operates.

6. the implementation method of a virtual machine cluster system, it is characterized in that, this virtual machine cluster system is structured on a plurality of physical servers, move virtual machines so that a plurality of virtual computing nodes to be provided at those physical servers, those virtual computing nodes are divided into a management node and a plurality of computing node, virtual computing nodes all in the whole virtual machine cluster system all use same disk mirroring, wherein:

Create host name and the dynamic IP addressing of a management holder to record each computing node at this management node, and according to those host name and dynamic IP addressing whole virtual machine cluster system is managed; The physical server that the expansion servers of this management node will newly access this virtual machine cluster system is configured to move the virtualization services supplier VSP of computing node;

Wherein, those computing nodes are used for finishing the calculation task of this virtual machine cluster system.

7. method according to claim 6 is characterized in that:

Those physical servers adopt the Copy on write form of Qemu that the disk mirroring of virtual machine is provided.

8. method according to claim 6 is characterized in that:

Those physical servers use the network file system(NFS) shared disk.

9. method according to claim 6 is characterized in that:

Setting up heartbeat between this management node and those computing nodes is connected.

Images