Distributed Computing's Disadvantages

All the computers on the Internet may be powerful when combined, but there needs to be something to combine and coordinate all of them to work towards one goal. Server computers are still needed to distribute the pieces of data and collect the results from participating clients. Collecting Internet users and having them sign up for a distributed computing program is not an automatic process: often times, organizations need to attract users with offers and incentives, from the aforementioned screensaver to cash payments.

Distributed computing must be flexible as well because of the nature of the Internet. Many people who use the Internet still do so using a regular modem, so communication with participating computers can be a slow. For this reason, data must take up as little bandwidth and as little disk space as possible. The data must also be independent of each other.

Fig. 1: an n-body simulation running on computers

Certain calculation processes depend on many variables and thus the computer's processor chip needs to consult all the other computers to solve these calculations. Consider an n-body simulation, which calculates the effects of a force field on atoms, whether it be a gas or a fluid. One processor calculates the path of one of these atoms. These calculations depend on the forces acting on it from other atoms, which are calculated by the other processors. Thus, the processor relies on its fellow processors and needs to communicate with them. (Hayes) Lack of effective communication would compromise the speed of the calculation process. Since most home computers are not continuously online, the Internet does not offer this luxury to the computer's processor.

Because the Internet is out in the open, security is a major issue and concerns both the organization using distributed computing and the user whose computer is doing the work. The organization needs to be able to trust the data results that the user's computer provides; several problems in the past include the "tweaking" of the software to report a faster processing speed to malicious data with fake alerts, which needed to be recalculated again. The user must also be able to trust the organization — how will the user know whether the organization is really processing, say, prime numbers? — and the organization's software — how will the user know whether the software isn't digging through his files?

When dealing with a local distributed network — where the computers are within a contained area, such as a room — the cost of maintenance can skyrocket because each computer has its own share of problems and errors. Unless these computers are used for something else, say, as workstations at a library, this type of distributed computing isn't cost-effective.

©2000 Team DC (Thinkquest Team C007645). Hosted by ThinkQuest.