Server consolidation is important in situations where a sequence of database tenants needto be allocated (hosted) dynamically on a minimum number of cloud server machines. Givena tenant’s load defined by the amount of resources that the tenant requires and a service-level-agreement (SLA) between the tenant customer and the cloud service provider, resource costsavings can be achieved by consolidating multiple database tenants on server machines. Ad-ditionally, in realistic settings, server machines might fail causing their tenants to become un-available. To address this, service providers place multiple replicas of each tenant on differentservers and reserve extra capacity to ensure that tenant failover will not result in overload onany remaining server. The focus of this thesis is on providing effective strategies for placingtenants on server machines so that the SLA requirements are met in the presence of failure ofone or more servers. We propose the Cube-Fit (CUBEFIT ) algorithm for multitenant databaseserver consolidation that saves resource costs by utilizing fewer servers than existing approachesfor analytical workloads. Additionally, unlike existing consolidation algorithms, CUBEFIT cantolerate multiple server failures while ensuring that no server becomes overloaded. We provideextensive theoretical analysis and experimental evaluation of CUBEFIT. We show that comparedto existing algorithms, the average case and worst case behavior of CUBEFIT is superior and thatCUBEFIT produces near-optimal tenant allocation when the number of tenants is large. Throughevaluation and deployment on a cluster of up to 73 machines as well as through simulation stud-ies, we experimentally demonstrate the efficacy of CUBEFIT in practical settings.
展开▼
