¶ … ability of the system to perform its function even in the presence of failures," fault tolerance is a critical component of effective cloud computing systems (Jhawar & Piuri, 2013, p. 3). To improve fault tolerance, a system must be designed to anticipate errors, which is why fault tolerance can be considered a form of risk management. Usually, cloud systems with solid fault tolerance can handle functioning at reduced capacity in order to maintain critical systems components defined by the core stakeholders or end users (Amin, Sethi & Singh, 2015). Because each cloud system will be differently constructed, fault tolerance architecture improvements must reflect the specific features and needs of the individual system.
Generally speaking, four main types of clouds inform the methodologies of fault tolerant designs. Those four types of clouds include public, private, hybrid, and community clouds. All of these types will offer various services loosely grouped into three categories: software as service, platform as service, and infrastructure as service. Depending on the services and type of cloud in consideration, fault tolerance may be reactive, proactive, or both. Moreover, the means by which fault tolerance is designed and measured for its effectiveness varies depending on needs for response times, throughput, and...
Types of faults may also be taken into consideration when designing effective fault tolerant systems; network faults, physical faults, media faults, processor, process and service expiry faults are all possible and can be classified as being intermittent, permanent, or transient in nature (Saikia & Devi, 2014). Improving fault tolerance in the cloud necessitates an understanding of the specifications of the system and its components.
Reactive fault tolerance can be used when failures are relatively regular and predictable. Types of reactive fault tolerance include job migration, in which a task can be physically designated to a predetermined backup machine using HA proxy or other proven methods (Bala & Chana, 2012). Check pointing is an effective reactive technique particularly useful for "long running and big applications," (Saikia & Devi, 2014, p. 4). Bilal, et al. (2015) note that fault tolerant systems can use a roll-forward or roll-back method in order to revert to a system state prior to the error. Checkpoint mechanisms, although not always cost-effective, can be particularly useful fault tolerance methods. Various types of checkpoint mechanisms include checkpoint placement schemes, in which checkpoints are inserted…
