Grid computing has been of tremendous use to companies, both large and small scale, for quite a few years now. Grid computing, in short is a simple and easy way to carry out computing jobs like database queries with the help of a set of resources instead of a single central core. Silos or huge SMP boxes were used in the past and even now some companies use the SMP boxes but they are not economical and also not very effective. With the advent of grid computing the load on the server gets distributed evenly to small server components and jobs are carried out easily and efficiently.
Here are some advantages and disadvantages of grid computing.
Advantages
The need for huge SMP servers is eliminated and instead small servers which are cost efficient are used. The resulting work can be analyzed and the job made simpler.
Many resources are not utilized efficiently except during business hours and with the grid computing these idle computers or servers are put to good use. If there is a fault in one of the computers in a grid, the job is taken over by other servers or computers in the grid and thus the work does not come to a stand still. Jobs are programmed to restart automatically in case of system faults.
Grid software like Platform LSF, Data Synapse, Sun Grid Engine, Condor and UnivaUD are used to manage policies. Resources are surveyed and job reports are managed on the basis of internal policy engines. The grid computing has the ability to scale well. If more computing resources are required then they can be easily acquired by connecting additional servers or computers which can be detached as easily, if not needed. Upgrading is an easy task without the interruption of the ongoing work. One of the resources can be used for upgrading while others take up the work and see to its completion. Parallel speeding performance can be seen in the completion of the various jobs. With the use of MPI messages can be transmitted within the various resources easily.
Disadvantages
In case of applications that do not make use of the MPI model, reverting to the SMP may be needed. High speed interconnection between the computer resources like a gigabit Ethernet is needed and for MPI specific applications Infiband may be required. Tweaking of the application may be required in a few cases to get the full effect of the computing.
The process of licensing used in various servers at a time is not suitable for certain type of applications. The manufacturers have begun to be more adaptable in such an environment.
With the many advantages associated with the grid computing, it is clear that the benefits accrued are more and a proper knowledge about the computing will take care of the disadvantages realized. Areas like bioinformatics, oil and drilling and various financial applications have started to enjoy the benefits and it will not be long before all fields apply grid computing with resounding success.
