Server parallel computing architecture solution
Time:05/28/2018 Hits:6725
A clustered computer is a set of networked PCs or workstations that provide much higher computing performance than a single PC (node). To say that it is a computer is actually for the provided service, or is logically speaking.
Background introduction
A clustered computer is a set of networked PCs or workstations that provide much higher computing performance than a single PC (node). To say that it is a computer is actually for the provided service, or is logically speaking. The main advantage of cluster computers is cost. Under the same computing power, the cost is only one-tenth that of traditional mainframes. In the global computer TOP500, the proportion of cluster computers is increasing, and the proportion in the past two years has exceeded 80%. . Especially for the application of SMEs, experimental research and teaching are very competitive.
The cluster system mainly solves the following problems:
1. High reliability (HA). With the cluster management software, when the primary server fails, the backup server can automatically take over the work of the primary server and switch over in time to achieve uninterrupted service to users.
2. High Performance Computing (HP). Make full use of the resources of each computer in the cluster to achieve parallel processing of complex operations, usually used in scientific computing, such as genetic analysis, physical and chemical analysis.
3. Load balancing. That is, the load pressure is reasonably distributed to each computer in the cluster according to an algorithm to relieve the pressure on the host server and reduce the hardware and software requirements for the host server. In practical applications, the most common situation is to use clusters to solve the problem of load balancing, such as providing WWW services. Here mainly shows how to use LVS (Linux Virtial Server) to achieve a practical WWW load balancing cluster system operating system on the cluster computer, in the TOP500 now the main operating system has a figure, but the most widely used is the LINUX operating system. This is of interest to Linux's open source code first, which makes it highly configurable. It can be modified according to different hardware and application environments. Secondly, Linux has high stability and there is no lack of commercial support. In order to reduce the difficulty of program development and improve the portability of the program, the cluster-based computing environment based on LINUX is mainly realized by configuring the MPI (Message Passing Interface) environment. This is also a more mature application mode.
Scheme topology
Explanation:
Regarding server parallel computing, the server we use is as follows:
1, the main node server: 4-way server 4CPU 256GB memory 2T storage
2. Node Server: Dual Server 2 CPU 256GB Memory 1T Storage
advantage:
1, 4-way server products are located in high-performance operations, databases and other important and need stable machines. Can play its computing performance outstandingly.
2. Two-way server is used as a node server, with 256G memory and 2 CPUs, providing the computing power of each node.
The configuration of the cluster environment mainly includes the following parts:
1, the installation and configuration of parallel computing environment
2. Compute passwordless access between nodes
3, the configuration of the shared file system