Degrees of tenant isolation for cloud-hosted software services: a cross-case analysis

摘要: A challenge, when implementing multi-tenancy in a cloud-hosted software service, is how to ensure that the performance and resource consumption of one tenant does not adversely affect other tenants. Software designers architects must achieve an optimal degree isolation for their chosen application requirements. The objective this research reveal trade-offs, commonalities, differences be considered required isolation. This uses cross-case analysis selected open source engineering tools empirically evaluate varying degrees between Our reveals five commonalities across case studies: disk space reduction, use locking, low cloud consumption, customization plug-in architecture, choice pattern. Two these common factors compromise reduced there no strategy reduce architecture adopted. In contrast, improves careful consideration given handle high workload, locking data processes used prevent clashes multiple tenants selection appropriate also revealed study differences: size generated data, sensitivity workload changes, effect process, client latency bandwidth, type process. impaired, our results, by large certain processes, or fluctuating latency, bandwidth transferring files repositories. Additionally, provides novel explanatory framework (i) mapping different development resources layers stack; (ii) explaining trade-offs consider affecting (i.e. sharing, number users/requests, customizability, scope control stack business constraints) multi-tenant services. suggests have pay attention we identify