There is a big difference between centralized and distributing operating systems. Centralized systems basically have processors that share a block of memory that is needed to send and receive data between the computers that are connected to. This is different form distributed operating systems. It takes a lot for it to work the way it supposed to, but sometimes there are instances that will hinder distributed systems to improperly operate. There are four types of failures in distributed systems. The first type of failure in a distributed operating system is communication failure.
In distributed systems, each computer has its own processor and each have its own block of memory that is private to that computer. These computers pass messages to each one machine or to all of the machines that is connected through the distributed system. Some of the tasks that can be performed are email, transferring information, and peer-to-peer networking. A problem can arise when there is an error in communication. For example, computer #1 can send messages to computer #2, but it cannot receive them from the same machine.
This can be caused by a delayed response between the processors and the servers trying to send and receive messages across the distributed system and the status of the servers and routers, because any network devices that are not in good shape could result in loss of connection, data, slow performance form the server, or access to the system. Speaking of network devices, the second kind of failure of a distributed operating system is in the event when any kind of component that is faulty can result in the entire computer to crash and ossible disabling of the distributing operating system.
Some device inside of the computer is causing the processor to lose contact with the network, whether it may be worn out, old, or not compatible with the computer. Failed disk drives, faulty processors, chip, and other software problems can result in this type of failure. When this occurs, a fail-stop takes place. As the third type of failure in distributing operating systems, fail-stop is when the processor will stop all computing operations before incorrect processes will be performed.
The processor could be performing very slowly in one machine and the other processor will notice it, but when that same processor starts to regain regularity, the other processors will not be able to work properly. The fourth kind of failure in distributing operating systems is the Byzantine failure. This is a burden that takes place randomly that leaves the distributed system in an erratic method. Instead of the processor ceasing operation when errors show up like a fail-stop, it will continue to function.
The defective processor in one computer will continue send messages to the other processors, as if everything is working properly. This will result in incorrect messages passed between computers and duplicate file transfers from the servers. The two types of failures that can occur in both centralized and distributed operating systems are faulty component issues and the fail-stop failures. If there is a piece of equipment, hardware or software that does not function to the best of its capabilities the network communication and operating system will ultimately fail and shut down.
Fail-stop failures completely stop all actions before all processes will see the unwanted operation. In order to attempt to fix these errors some actions must take place. The programmer must physically back up all important data every time that they update their programs and data. Also, programmers must make sure each essential piece of hardware and updated software is compatible with the variety of objectives that they want to achieve. This could help with the fail-stop failure and the faulty component failure.