Failures

POS/355
August 26, 2013
UOPX

Failures
Distributed systems emerged recently in the world of computers. A distributed system is an application of independent computers that appear to work as a coherent system to its users. The advantages of distributed systems consist of developing the ability to continually to open interactions with other components to accommodate a number of computers and users. Thus, stating that a stand-alone system is not as powerful as a distributed system that has the combined capabilities of distributed components. This type of system does have its complications and is difficult to maintain complex interactions continual between running components. Problems do arise because distributed systems are not without its failures. Four types of failures will characterize and the solutions to two of these failures will address on how to fix such problems.
Before constructing a distributed system reliable one must consider fault tolerance, availability, reliability, scalability, performance, and security. Fault tolerance means that the system continues to operate in the event of internal or external system failure to prevent data loss or other issues. Availability needed to restore operations to resume procedure with components has failed to perform. For the system to run over a long period without any errors is need and known as reliability. To remain scalable means to operate correctly on a large scale. Performance and security remains needed so the system can resist attacks against viruses and errors, so it can run in a timely manner. Distribution systems have their advantages and disadvantages.
The disadvantages are that there are more than just four types of failures. Resolving failures is important and typically fall into two main categories: hardware and software failures. Understanding the classification of the types of failures that may occur with distribution system helps pinpoint the error for repair. Knowing that distributed systems are known for failure here are the four types that chosen for understanding: Hardware failure, application failure, network failure, and synchronization failure. These four failures are very common on distributed systems, also critical to understand that failures happen to both a distributed system and a centralized system. Distributed systems are two or more computers linked through networks. The wLAN or WAN and have their own processors. A centralized system is a computer that runs independently. An example of this would happen on home laptop not part of a network.
Concerns for hardware failure was significant in the beginning of computer system history. Power consumption of small circuits, heating problems, and off-chip connection and writing remained improved over time through quality manufacturing techniques. If failure takes place meaning that lost information/data happens it can manage with the use of clustering. Using redundant shared disks or mechanisms for committing complete transactions is known as fault-tolerant cluster. Failover and failback are clusters using to switch an application or restore applications. Network failure occurs when a link fails to connect within a network. Using a service-oriented architecture to assist with the widespread use in enterprise systems stays fixed. Operating middle ware, database applications, and server availability hardware failure occur minimized.
When software on a computer is not working properly it is known as failure. Software failure has different subsets of failures that coexist in the distribution system. Among those types of failures happen to those addressed in the above passage. The third type of failure is application failure happens when an application fails to operate the way its suppose to or stops working altogether. The last type of failure is with data on different systems do not sync correctly with one another. Database applications, three-tier client/server architecture, and distributed message passing are unique ways to solutions for software failures like synchronization failure and application failure. Firewall remains used to protect software from errors. Centralized systems use LAN also. Users should not know that a machine is distributing and contains multiple CPU’s and to accomplish this is must have global protection schemes and a kernel on each machine responsible for the controlling of its own protocols. Distributed messaging passing meaning that a processor is willing to receive data to directing it to a buffer area and informing the message-passing module by a primitive. Caches are strategies for dealing with components of state cross data.
Both Client and servers must stand able to operate correctly with unresponsive senders and receivers. Latency remains valued to know the time between request of data and the start of data transfers. Limiting latency controls the amount of data shared and transferred. Sometimes data sent across a network is not sent or received. Communication failure could happen handled with client-server applications. Dependable communication means database servers, file servers, and network servers working together to map a way for applications to pass through.
Distributed systems and centralized systems both are design to have failures. It is how those failures occur carefully assisted and solved. Knowing this everyone should understand that systems happen created for failure. There are two types of failure: hardware and software. To truly have a reliable distributed system is to have the characteristics of fault-tolerance, availability, scalability, and security. Failure scenarios remain unlimited ways to identify how likely each failure may occur. Minimizing how much data is sent and secured is another way to have a reliable distributed system. Distribution systems are better than centralized systems when it approaches to a large scale of communication of information between users.

References
Krzyzanowski, P. (2003, 01 01). A taxonomy of distributed systems. Retrieved 08 20, 2013, from Rutgers University : http://www.cs.rutgers.edu/~pxk/rutgers/notes/content/01-intro.pdf
Stallings, W. (2012). Operating Systems: Internals and Design Principles, (7th ed.). Upper Saddle River, New Jersey: Prentice Hall.