What is the Difference Between Cluster Computing and Grid Computing?

Grid computing A problem solving method based on grid-based distributed computing. It uses computing resources in the network to solve a scientific or technical problem, which requires a large amount of computer processing power or access to a large amount of data. Cluster computing refers to a computer cluster that connects a group of loosely integrated computer software or hardware to work closely together to complete computing work. Grid computing and cluster computing are both to meet the needs of large computational requirements.

Cluster computing refers to a computer cluster that connects a group of loosely integrated computer software or hardware to work closely together to complete computing work. In a sense, they can be seen as a computer. A single computer in a cluster system is often called a node and is usually connected via a local area network, but there are other possible connections. Cluster computers are often used to improve the computing speed and / or reliability of a single computer. Cluster computers are generally much more cost-effective than single computers, such as workstations or supercomputers.

According to whether the architecture of the computers constituting the cluster system is the same, clusters can be divided into two types: homogeneous and heterogeneous. Cluster computers can be divided into high-availability (HA) clusters, load-balancing clusters, high-performance (HPC) clusters, and grid computing. computing).

(1) The main difference between a grid and a traditional cluster is that the grid is connected to a group of related and untrusted computers. It operates more like a computing utility than an independent computer. A grid typically supports more collections of different types of computers than a cluster.

(2) The grid is dynamic in nature, and the number of processors and resources contained in the cluster is usually static. On the grid, resources can appear dynamically, and resources can be added to or deleted from the grid as needed.

(3) Grids are inherently distributed on local networks, metropolitan area networks, or wide area networks. The grid can be distributed anywhere. The clusters are physically contained in the same place in one location, usually just LAN interconnection. Cluster interconnect technology can produce very low network latency, which can cause many problems if the clusters are far apart. Physical proximity and network delay limit the ability of the cluster to distribute geographically, and the grid can provide good high scalability due to its dynamic characteristics.

(4) The cluster only meets the growing demand by adding servers. However, the number of servers in the cluster, and the resulting cluster performance, is limited: interconnect network capacity. In other words, if we blindly want to increase the performance of the cluster computer by expanding its size, its cost performance will decrease accordingly, which means that we cannot expand the size of the cluster without limit. The grid virtualized an unprecedented supercomputer, which is not limited by the size, and has become the development direction of the next generation Internet.

(5) Cluster and grid computing are complementary. Many grids use clusters in resources they manage. In fact, the grid user may not know that his workload is executed on a remote cluster. Although there are many differences between grids and clusters, these differences make them a very important relationship, because clusters always have a place in the grid-specific problems usually require some tightly coupled processors to solve. However, with the development of network functions and bandwidth, problems that were previously difficult to solve with cluster computing can now be solved using grid computing technology. It is important to understand the balance between the inherent scalability of the grid and the performance benefits provided by the tightly coupled interconnect mechanism provided by the cluster.