It's a common faith that the Internet is a system of cloud computing. But the question remains how it is really possible for a system to handle such amount of huge data exchange at a time? It must be a real challenge for the team deployed at the cloud computing services to maintain the functioning given the fact that some regions of the world generate excessive data in real time. The problems of such extents are taken care of by load balancing.

Load balancing is the process in which the load of a particular data center in a cloud computing system is distributed almost equally to the available infrastructure. This way, the efficiency escalates by many notches while the facility doesn't suffer any infrastructural loss. The load balancing theory does not keep a record of the previous behavior of the centers and evaluates them on the present state. The load balancing algorithm is prepared keeping a number of factors in consideration - tentative calculation of the load, the consistency of different systems, the interconnections and compatibility of the nodes etc. The load could be any sort pertaining to the processing unit, the storage issue or the performance of applications.

Of the major responsibilities that load balancing is accountable for, system stability and to integrate further system modifications are crucial. This is the algorithm that is responsible for the continuation of applications in case a certain infrastructure of the cloud fails. The load balancers instantly transfer the process to other nodes of the cloud, thus not at all compromising the functioning.

The load balancing can be listed in a number of variants depending upon the process initiated namely Sender initiated, Receiver initiated and Symmetric. While the first two are self-explanatory the third is a combination of the couple. However, with reference to the present state of the process, Symmetric has been further divided into two options named Static and Dynamic.

Static, as the name suggests is dependent on the previous records of the process and does not take cognizance of the present state while dynamic is just the opposite and requires no prior records. The dynamic algorithm is further differentiated into two needful segments - distributed and non-distributed systems. Distributed is more of use when the nodes of the system act individually with minimal linkage amongst them. On the other hand, non-distributed is efficient for a centralized cluster of nodes. The central nodes are advised for large cloud clusters for if it crashes at any time, the whole system is reduced to utter uselessness. All of these are responsible for the user specific functionality of the cloud.

The above information makes it very much clear that load balancing is a crucial component and cannot be done away with. The performance and stability of the cloud is dependent much on this balancing process. Much of research and analysis is undergoing in this segment for betterment of the cloud computing.