Grid computing enables aggregation of distributed resources located at multiple locations connected with each other to solve a complex problem. All the computers connected are linked to each other using a loosely connected low-speed networks.
The difference between conventional high-performance computing and grid computing is that in the grid, each node is set to perform completely different tasks or application. The computers connected are heterogeneous in nature and are distributed geographically at different locations.
The servers and computers connected runs independent tasks and can also be connected directly or through scheduling systems. Grid computing which emerged in the 1990s was merely an evolution of cluster computing. Before understanding what is grid computing and how it is different from cluster computing and cloud computing, it is important for you to understand what are grids and clusters.
- 1 What is a Grid in Grid Computing?
- 2 Grid Computing Overview
- 3 How Grid Computing Works
- 4 Grid Computing Architecture
- 5 Grid Computing Concerns
- 6 Types of Grids
- 7 Difference Between Data Grids and CPU Scavenging Grids
- 8 Applications/Projects of Grid Computing
What is a Grid in Grid Computing?
A grid can be considered as a connection of parallel nodes to form a computer cluster that runs on an operating system or free software. Grids can be considered as distributed systems having non-interactive workloads involving a large number of files. Grids are mainly constructed through general purpose grid middleware software libraries whose size can be quite large than a cluster.
In some of the application grid computing can be explained in the form of parallel computing that relies on complete computers (having CPU, memory, power supplies, NIC card etc) which are connected to private or public networks through network interface card such as Ethernet.
Grid Computing Overview
The basic idea behind grid computing is to utilize the CPU and storage of many different computers residing at a different geographical location. Most of the application which is using grid computing for problem solving have no time dependency and the programs use an idle power of computers deployed across different countries which is also known as cycle-scavenging.
The size of the grid can vary from one organization to several many organizations. The notion of confined nodes is also known as intra-nodes cooperation whereas on another hand wider grids can be referred as inter-nodes cooperation.
Comparison of Grids and Conventional Computers
The question which must be not clear with most of you is what exactly is the difference between grid computing, distributed computing and cloud computing?
In the case of grid computing and distributed computing, both use parallel computing which completely relies on individual computers with primary memory and storage connected to the network through Network Interface Card. There is also some difference in processing and deployment. In the case of supercomputers, the cost can be much higher as compared to distributed network.
It is also much difficult to write programs which can be deployed individually on a supercomputer. A single computer can never be much efficient as a combined grid of computers. The advantage here is that the single program can be deployed at different locations and if debug at a single machine than the instance of the program running on different computers is not affected.
How Grid Computing Works
By now you must be clear with what is grid computing and its overview. Let’s try to understand how it works. Grid computing working is almost similar to that of distributed computing or it is a special kind of distributed computing. In distributes computing, all the computers connected to same network share one or more resources but in grid computing, every resource is shared making the whole system into a powerful supercomputer.
One of the advantages of grid computing system is that it can be simple as a collection of similar computers and at the same time as much complex as an inter-networked system comprised of different computers platforms. Along with that, each computer connected to the network have access to enormous processing power and memory.
The instance of the program is created and deployed at different machines located at other geographical locations providing an increase in computational power and less time to solve the same problem.
Grid Computing Architecture
Grid Computing Concerns
While you are connecting two or more computers to each other there are some points to be considered before connecting such as:
- Is your data secure?
- How will you be keeping your personal data private?
- How will the system be protected from malicious hackers?
- Who will be accessing your shared resources and how to keep track of each user accessing the information.
The solution can be the use of middleware. There are protocols for grid computing to facilitate communication between computers.
Types of Grids
Grids can be classified into four categories depending on their usage and usefulness. They are:
- Computational Grids
- Data Grids
- On-Demand Grids
- Management Grids
Difference Between Data Grids and CPU Scavenging Grids
Data grids consist of systems designed for handling large distributed data sets which are used for data management and control data sharing. These types of systems use middle software to create a digital library with dispersed file system.. An example of data grids is Southern California Earthquake Center.
CPU Scavenging Grids have a cycle-scavenging grid system which moves projects from one computer to another according to the needs.An example of this type of grid is extraterrestrial intelligence computation comprising of more than three million computers.
Applications/Projects of Grid Computing
You can find many applications of grid computing but all of them does not fit part into the definition of grid computing. The Search Extraterrestrial Intelligence (SETI) project is the earliest grid computing system which gained much popular attention towards itself.
The mission was to analyze data which was gathered by radio telescopes in order to get evidence for intelligent alien communications. As the information which was collected by telescopes was too much for the single computer to be analyzed.
The SETI@home project was started in which million of PCs run a search program on a segmented piece of data collected by radio telescope and date for completion was not set. Grid computing can be seen in places where there is a requirement of high computational power.