Online Failure Forecast For Fault Tolerant Data Stream Processing

Asymmetric clustering: one host runs the database application while the other host simply monitors it. If the server fails, the monitoring host becomes the active server. This is appropriate for providing redundancy. However, it does not utilize the potential processing power of both hosts.

keep track of its state.If the first node fails, the second node takes up the operations from where the

The reliability of a system can be improved through the introduction of redundancy in the system. Some of the examples of redundancy in the operation are as below:

First off to start the assignment only requires writing about four different types of failures that can happen on a distributed system, however there are many more than just four types of failures that can happen and they are all important to learn about if you are going to work with a distributed system so that you know how to deal with and handle each one of them.

MPI: The fault-tolerance mechanism in MPI depends either on handling failure in the application itself or implementing regular checkpoint files.

A goal of redundant topologies is to eliminate network down time caused by a single point of failure. All networks need redundancy for enhanced reliability. Network reliability is achieved through reliable equipment and network designs that are tolerant to failures and faults. Networks should be designed to reconverge rapidly so that the fault is bypassed.

Recommendation: To recover data packets lost in transit; the data loss caused by a computer/server crash can be prevented using a solid backup plan. Data files can be replicated over a remote server as well as on different computers on a network.

This system ignores the failure process and takes successful process to the higher level within the fixed time period.

IEEE Std 982.1-2005 (Revision of IEEE Std 982.1-1988) - IEEE Standard Dictionary of Measures of the Software Aspects of Dependability

Equipment Failure - A hard drive fails in the network server. If you built fault tolerance into your network. A fault tolerant system has an online redundant component ready to go; if the primary dies, the system switches

An IT disaster occurs when a business’s mission- critical system is not available. An IT disaster may be nature- or human- induced. Natural disaster including fire, hurricane, flood, earthquake, tornado etc. can damage the IT physical infrastructures. Fire is reportedly one of the most common catastrophic events that can interrupts a business operation (....) Human- induced disaster include terrorist attacks, deliberate criminal activities etc. An IT disaster can have serious consequences to a business if not being addressed timely and correctly. Apart from loss of existing data,

The N-version software concept attempts to parallel the traditional hardware fault tolerance concept of N-way redundant hardware. In an N-version software system, each module is made with up to N different implementations. Each variant accomplishes the same task, but hopefully in a different way. Each version then submits its answer to voter or decider which determines the correct answer, (hopefully, all versions were the same and correct,) and returns that as the result of the module. This system can hopefully overcome the design faults present in most software by relying upon the design diversity concept. An important distinction in N-version software is the fact that the system could include multiple types of hardware using multiple versions of software. The goal is to increase the diversity in order to avoid common mode failures. Using N-version software, it is encouraged that each different version be implemented in as diverse a manner as possible, including different tool sets, different programming languages, and possibly different environments. The various development groups must have as little interaction related to the programming between them as possible. N-version software can only be successful and successfully tolerate faults if the required design diversity is met.

And also it provide fault tolerance platform is capa-ble of executing messages using Message Passing Interface In order to hold fault tolerance in a Distri-buted Multithread.

Data stream is the continuous flow or stream of data and if it is not processed immediately or stored, then it is lost forever. The arrival of data is so quick that it is not feasible to store it completely in storage, and then perform several queries. That’s why we require some technique or method for processing streams which involve summarization of the stream in some way. This technique generalizes to approximating various kinds of Data Stream Mining Methods, some of which we will discuss in section 2. We will discuss the challenges in section 3.

This subsection indicates the accompanying necessities connected with the dependability (e.g., interim between failures, number of failures for every unit time) of the system.