Flexible fault tolerance in cloud through replicated cooperative resource group

Abstract

Listen

Translate article

Fault tolerance (FT) is a highly focused issue in cloud due to high reliability requirements especially for delay sensitive tasks. Numerous FT frameworks in cloud have been proposed in literature. By and large, proactive and reactive FT approaches are followed in the existing frameworks. Irrespective of any of the applied approaches, FT implementation remains service provider centric in the existing frameworks. Cloud users are not given any control over deciding the level of FT. However, cloud follows the service orientation model with greater flexibility in using the computing services on demand basis and in pay-per-use manner. The user’s control in deciding the level of FT while executing their tasks will further reinforce the model of service orientation in cloud. Therefore, this paper proposes a flexible fault tolerance framework (FFTF) in cloud. FFTF provision the users to categorize their tasks as premium/standard/economy to implement the corresponding level of fault tolerance (FT). To implement a FT level, user task is executed on a cooperative resource group in cloud. Flexibility of FFTF is endorsed by maintaining the performance parameters of deadline guarantee ratio and average task delay for different task categories. The performance of FFTF is analyzed and compared through extensive simulation experiments on artificial and real workload in term of its FT capability, resource consumption and utilization with an existing FT framework.