Fault tolerating multi-tenant service-based systems with dynamic quality

摘要

The rapid development of multi-tenant service-based systems (SBSs) is mainly due to the popularity of cloud computing. SBS combines the existing services in the form of business processes to meet the needs of multiple tenants. There are two kinds of services: functional and non-functional. Multiple services can have the same functionality while changing their non-functional properties, or QoS (Quality of Service). QoS value is an important criterion for service selection or recommendation. In the existing work, QoS value in service selection or recommendation is regarded as a constant. However, in the dynamic and volatile cloud environment, the performance of services may fluctuate due to various factors (e.g. techniques of service studied, network conditions, etc.). A constant value does not reflect the consistency of the service. This paper proposes FT4MTS-D (fault-tolerant strategies for multi-tenant service-based systems with dynamic quality), a novel method that evaluates the criticality of dynamic cost-effective fault tolerance strategies for multi-tenant service-based systems. First, the consistency of each given service is evaluated based on the feature vector of the cloud model. Second, this paper measures the consistency of given service by the sensitivity of probability degradation of an SBS in comparison to the probability degradation of that service. The goal is to find the critical services in the SBS and then determine the optimal fault-tolerant strategy for the given service. The experimental results show that FT4MTS-D has obvious advantages compared with several existing representative methods.