Efficient Fault Tolerance on Cloud Environments

2. Fault Tolerance For Cloud Environments

Fault Tolerance aims to ensure systems are able to deliver in case of one of more failures of the unit’s components. Fault Tolerance (Anjali et al., 2016) is system resource availability and reliability not being affected in case any of the preceding component or execution devices (Mohammed et al., 2016) failing or there are multiple failures for the hosted application system or infrastructure devices (Zhang et al., 2011). Usually systems, devices or resources are often over provisioned or purposely underutilized to ensure even if the application performance might be affected during an outage, the systems continue to perform possibly at a reduced level, rather than failing completely within predictable and acceptable bounds. Fault tolerance is mostly implemented in high-availability life-critical system environments. Providing fault tolerant design (Patra et al., 2013) for each and every single component is however not an effective solution. The associated redundancy and over provisioning brings a number of parasitic penalties: increase in weight, cost, power, size, consumption, as well as time to design, verify and test before delivering the service. The following options are taken into account when determining how and why the computing components should be fault tolerant: