Abstract: Resources in an Infrastructure-as-a-Service (IaaS) system are upgraded in an iterative process. In response to an upgrade request indicating requested changes to a current configuration of the system, one or more graph representations of the current configuration and the requested changes are created. The graph representations include a control graph which has vertices representing resource groups, and edges representing dependences among the resource groups. A batch of resource groups is identified to be upgraded in a current iteration based on the dependencies and Service Level Agreement (SLA) requirements including availability and elasticity of the system. Upgrade operations are executed on the identified batch using selected upgrade methods which handle potential incompatibilties during transition of system configurations. The graph representations are updated to include any new requested changes and recovery operations in response to feedback of failed upgrade operations.

Abstract: Resources in an Infrastructure-as-a-Service (IaaS) system are upgraded in an iterative process. In response to an upgrade request indicating requested changes to a current configuration of the system, one or more graph representations of the current configuration and the requested changes are created. The graph representations include a control graph which has vertices representing resource groups, and edges representing dependences among the resource groups. A batch of resource groups is identified to be upgraded in a current iteration based on the dependencies and Service Level Agreement (SLA) requirements including availability and elasticity of the system. Upgrade operations are executed on the identified batch using selected upgrade methods which handle potential incompatibilties during transition of system configurations. The graph representations are updated to include any new requested changes and recovery operations in response to feedback of failed upgrade operations.

Abstract: A system is upgraded by rolling upgrade with dynamic batch sizes in multiple iterations. The system includes multiple hosts that host multiple resources. The rolling upgrade calculates a number of hosts to upgrade and a number of resources to upgrade for each iteration. The calculation subtracts a reserved number of hosts from a capacity of the system to account for potential failover and potential resource scaling operations during the iteration. Based on the calculated number of hosts to upgrade, it is determined whether the system has a capacity for upgrade. If the capacity exists, the calculated number of hosts to upgrade and the calculated number of resources to upgrade are upgraded, while accepting resource scaling requests during the iteration.

Abstract: A system is upgraded by rolling upgrade with dynamic batch sizes in multiple iterations. The system includes multiple hosts that host multiple resources. The rolling upgrade calculates a number of hosts to upgrade and a number of resources to upgrade for each iteration. The calculation subtracts a reserved number of hosts from a capacity of the system to account for potential failover and potential resource scaling operations during the iteration. Based on the calculated number of hosts to upgrade, it is determined whether the system has a capacity for upgrade. If the capacity exists, the calculated number of hosts to upgrade and the calculated number of resources to upgrade are upgraded, while accepting resource scaling requests during the iteration.