The Grid Computing is understood like the union of several administrative controls, with the purpose of getting a common end. In other words, it is a net that contains great quantity of archives interconnected among them, although this comes true of non-interactive way. The main characteristic that tells from this type of models is precisely that these tend to be less matched, a little dispersed and inclusively at times they can rate themselves of heterogeneous.
Often the Grid are used to accomplish several tasks at the same time, however, also they can take care of only a specific subject. It comes to be very simple to work with a suchlike structure, since most of programs find themselves organized as if they were within a library.
The applications that they find inside a Grid, are in reality little parts of the entire thing. For instance, if we made out an analogy of this structure and compare it directly with the human body, we would realize that every one of the above-mentioned applications would represent a little part of a muscle. While the remaining part resides in external computers. This aspect contrasts clearly with what commonly is acquainted like a supercomputer, because those apparatuses find themselves connected to several processors.
So that this gets explained of better way, from now on we will make a comparison between the Grid and the Supercomputers.
As we already said in previous paragraphs, the Grid are a particular class of computers that depend directly on other ones, which in turn provide it with all necessary tools for their correct functioning, connected to the intervening net Ethernet (CPU, storage, infrastructure of net, energy sources etcetera). This differs from the supercomputers, which find themselves connected in a terminal of local high-speed connection.
The first competitive advantage that offer the Grid Computing, is that each part can be acquired of individual way with the intention of later on being assembled in a team of bigger size. This perhaps correspond to the tendency that at the present time the personalized computers are created based on principles of efficacy - efficiency, but most of all looking for the low cost. But unfortunately not everything is good, since neither the processors used in these teams, neither the storage space have high-speed connections. In spite of this, the Grids come to be of great help when data do not call to be concentrated in a single place.
Just the same, we can find differences between the supercomputers and the Grids at the time of their establishment and programming. On the one hand, can come extremely expensive to develop programs lead to the supercomputers, since these generally are created to carry on a specific task or they have big patterns of personalization, thing that certainly makes difficult your operation.
Another important characteristic that have the Grids is that they can get constituted from different users or different organizations, as the occasion requires. This is known as: Administrative multiple domains. This structure facilitates most of all, the commercial aspect since it makes easy the interaction of voluntary way between individuals or companionships connected inside a common net. Nevertheless, a disadvantage of this model could be perhaps the fact that the performance of these machines, at the time of accomplishing complex calculations, it would not be precisely the most reliable.
For this reason, programmers should make sure to count on effective tools that they allow detecting and to correct these errors of fast and efficient way. If they do not accomplish it that way, teams can suffer information-technology attacks that they trigger in the obtaining of erroneous results. At the same time if this came to pass, what it would make was checking the same result in two different computers, in that way being completely sure of their veracity.
The aspects of security and performance can affect strongly at the time of taking a decision on which one is the structure that made more suitable the business necessity, either an internal mechanism an external mechanism headed by a net of no contentious users(supercomputers) (Grid). In most cases, customers trust that servers not saturate themselves. However, this can become very subjective since teams can find themselves in a state of vulnerability which at any time generates holes as to the internal security of the system. In order to avoid the information-technology attacks, a lot of companies have chosen to use virtual machines.
For their part, other organizations use personalized software, which allows them having once most of the aspects related with security were controlled. Even at times they come to designate complete areas to keep an eye on their information-technology surroundings.
Fuente: Cloud Computing
Fuente: Cloud Computing