Shulou(Shulou.com)06/02 Report--

Some of the contents of this article come from the following articles:

Http://soft.chinabyte.com/25/13169025.shtml

Http://www.linuxvirtualserver.org/zh/lvs1.html

The index of this article

What is a cluster

The characteristics of clusters

Advantages of clusters

Cluster classification

1 what is a cluster

Computer cluster, referred to as "cluster", is a kind of computer system, which works closely together to complete computing work through a set of loosely integrated computer software and / or hardware. In a sense, they can be regarded as a computer. A single computer in a cluster system is often called a node and is usually connected through a local area network, but there are other possible ways to connect. Cluster computers are often used to improve the computing speed and / or reliability of a single computer. In general, cluster computers are much more cost-effective than individual computers, workstations or supercomputers.

A cluster is a group of independent computers that form a computer system through a high-speed network, and each cluster node is an independent server running its own process. For network users, the back end of the website is a single system, which works together to provide users with system resources and system services. Combine a network connection into a combination to accomplish a task together. Common cluster results are shown below:

2 the characteristics of the cluster

1) High performance (performance)

Some require strong computing power, such as weather forecasts, nuclear tests and so on. This is not something that a few computers can handle. It takes thousands of sets to finish the job together.

The workload of network services is usually a large number of independent tasks, which can achieve high overall performance by dividing and conquering a group of servers.

2) Price effectiveness (cost performance)

Usually a set of system cluster architecture requires only a few or dozens of server hosts, which is more cost-effective than millions of dedicated supercomputers.

The PC servers or RISC servers and standard network equipment that make up the cluster system have the highest performance / price ratio because of low cost and low cost due to large-scale production. If the overall performance increases linearly with the increase of the number of nodes, the performance / price ratio of the system is close to that of the PC server. Therefore, this loosely coupled architecture has a better performance / price ratio than tightly coupled multiprocessor systems.

3) scalability

The number of nodes in the cluster system can grow to thousands or even tens of thousands, and its scalability is much higher than that of a single supercomputer.

When the server load pressure increases, the system can be expanded to meet the demand without reducing the quality of service.

4) High availability

There is redundancy in both hardware and software. High availability can be achieved by detecting the faults of software and hardware, shielding the faults and providing services by the surviving nodes. Even if part of the hardware and software fails, the service of the whole system must run 24 hours a day. When a module is found to fail, the services provided on that module are migrated to other modules. Ideally, this migration is immediate and automatic.

3 advantages of clusters

1) Transparency

If a part of the server goes down and the business is not affected, the degree of coupling is not so high and the dependency is not so high. For example, if the NFS server is down, the others cannot be mounted, which is too dependent.

How to efficiently make a loosely coupled cluster system composed of multiple independent computers to form a virtual server; when the client application interacts with the cluster system, it is like interacting with a high-performance and highly available server. The client does not need to make any changes. The cut-in and cut-out of some servers will not interrupt the service, which is also transparent to users.

2) High performance

The number of visitors increases and can be easily expanded.

If the performance is close to linear acceleration, it is necessary to design a good architecture of software and hardware to eliminate the possible bottlenecks of the system. Load is dispatched to each server in a more balanced way.

3) manageability

The entire system can be physically large, making the cluster system manageable, just like managing a single image system. Ideally, the insertion of hardware and software modules can achieve Plug & Play.

4) programmability

On a cluster system, it is easy to develop applications, which portals will require.

4 Cluster classification

[1] load balancing Cluster (Load balancing clusters) is abbreviated to LBC

[2] High availability Cluster (High-availability clusters) is abbreviated to HAC

[3] High performance Computing Cluster (High-perfomance clusters) is referred to as HPC

[4] Grid Computing (Gridcomputing)

On the network, it is generally believed that there are only three of the first three. "load balancing" and "highly available clusters" are commonly used cluster architectures in the Internet industry, and we must master them. Let's talk about the first three clusters in detail.

@ @ | load balancer cluster |

Load balancing cluster provides enterprises with a more practical and cost-effective system architecture solution. The load balancing cluster distributes the load load of many customers' centralized access requests to the computer cluster as evenly as possible. Customer request load usually includes "application level processing load" and "network traffic load". Such a system is ideal for providing services to a large number of users using the same set of applications. Each node can bear a certain load of access requests, and can achieve dynamic distribution of access requests among nodes to achieve load balancing.

When load balancing is running, customer access requests are generally distributed to a group of back-end servers through one or more front-end load balancers, so as to achieve high performance and high availability of the whole system. Such a computer cluster is sometimes referred to as a server farm. Generally, high availability clusters and load balancing clusters use similar technologies, or have the characteristics of both high availability and load balancing.

| the role of load balancer cluster |:

Share access traffic (load balancing)

Maintain business continuity (high availability)

@ @ | High availability cluster |:

Generally speaking, when any node in the cluster fails, all tasks on the node are automatically transferred to other normal nodes, and this process does not affect the operation of the whole cluster and does not affect the provision of services.

Similarly, there are two or more identical nodes running in the cluster, when a master node fails, then other nodes as slave nodes will take over the tasks above the master node. The slave node can take over the resources of the master node (IP address, schema identity, etc.), and the user will not find that the object providing the service is transferred from the master node to the slave node.

| the role of high availability clusters |:

When one machine goes down and the other takes over.

The more commonly used high-availability cluster open source software is: keepalive,heardbeat

@ @ | High performance computing cluster |:

High-performance computing cluster uses different computing nodes that assign computing tasks to the cluster to improve computing power, so it is mainly used in the field of scientific computing. The popular HPC uses the Linux operating system and some other free software to complete parallel operations. This cluster configuration is often referred to as a Beowulf cluster. Such clusters usually run specific programs to take advantage of the parallelism of HPCcluster. Such programs generally apply specific runtime libraries, such as the MPI library designed for scientific computing.

HPC cluster is especially suitable for computing jobs in which a large amount of data communication occurs between computing nodes, such as the intermediate results of one node will affect the calculation results of other nodes.

@ @ | commonly used cluster hardware and software |:

Commonly used open source cluster software is: lvs,keepalived,haproxy,nginx,apache,heartbeat

The commonly used commercial cluster hardware are: F5, Netscaler, Radware.A10, etc.

The author's note

Thank you for reading here. If this article can make you gain anything, I will feel extremely honored. If I can't help you, I will send you a sentence: deliberately practice 10,000 hours, you will enter the flow, continue to break through, and you will become a master.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.