Abstract: A method for recovering a storage policy of a workload executing in a cluster of host servers that are managed by a first management appliance, wherein the host servers each include a local storage device, and the storage policy corresponds to storage objects of the workload, includes the steps of: in response to an instruction from the first management appliance, creating a first storage object of the workload according to the storage policy, wherein the instruction includes the storage policy; storing the first storage object and the storage policy in a shared storage device that is provisioned from the local storage devices of the host servers; and in response to a request from a second management appliance configured to manage the cluster of host servers, retrieving the storage policy from the shared storage device and transmitting the storage policy to the second management appliance.

Abstract: Techniques for automated datastore unavailability handling are provided. In one set of embodiments, a computer system can receive a request to bring a datastore offline and, in response to the request, identify one or more virtual machines (VMs) in a virtualized computing environment that have one or more virtual disks stored in the datastore. The computer system can then, for each of the one or more VMs, determine an action to be taken with respect to the VM in response to the datastore's unavailability and trigger execution of the action.

Abstract: The disclosure provides an approach for high-availability admission control. Embodiments include determining a number of slots present on the cluster of hosts. Embodiments include receiving an indication of a number of host failures to tolerate. Embodiments include determining a number of slots that are assigned to existing computing instances on the cluster of hosts. Embodiments include determining an available cluster capacity based on the number of slots present on the cluster of hosts, the number of host failures to tolerate, and the number of slots that are assigned to existing computing instances on the cluster of hosts. Embodiments include determining whether to admit a given computing instance to the cluster of hosts based on the available cluster capacity.

Abstract: The present disclosure is related to methods, systems, and machine-readable media for force provisioning virtual objects in degraded stretched clusters. A request to provision a virtual object by a stretched cluster according to a storage policy specified as part of the request can be received by a software defined data center (SDDC). The cluster can include a plurality of sites. An insufficiency of storage policy resources to satisfy the storage policy specified for the virtual object can be determined. The virtual object can be force provisioned responsive to determining storage policy resources sufficient to satisfy the storage policy at one of the plurality of sites.

Abstract: In certain embodiments, a computer system can create first and second pluralities of disk groups in a hyperconverged infrastructure (HCI) cluster, where each disk group in the first plurality has capacity storage devices of a first type and each disk group in the second plurality has capacity storage devices of a second type. The computer system can further tag each disk group in the first plurality with a first disk group tag, tag each disk group in the second plurality with a second disk group tag, and create a storage policy that includes a placement rule identifying the first disk group tag. Then, at a time of provisioning a virtual machine (VM) in the HCI cluster that is associated with the storage policy, the computer system can place the VM on one or more of the first plurality of disk groups in accordance with the placement rule identifying the first disk group tag.

Abstract: A method for storage management of an object among a plurality of storage devices of a datacenter is provided. The method, in response to receiving an input on a selection item presented through a UI, determines that a manual storage management of an object is selected. The method then receives a storage policy for storing the object. Based on the storage policy, the method defines a plurality of components for the object and determines whether a set of one or more storage resources is available for storing the plurality of components. When the method determines that the set is available, for each component, the method presents the set of storage resources, receives a selection of a storage resource in the set to store the component, and updates the set based on the policy and the selection before presenting the updated set to select from for storing a next component.

Abstract: An example method of placing a virtual machine (VM) in a cluster of hosts is described. Each of the hosts having a hypervisor managed by a virtualization management server for the cluster, the hosts separated into a plurality of nonuniform memory access (NUMA) domains. The method including: comparing a virtual central processing unit (vCPU) and memory configuration of the VM with physical NUMA topologies of the hosts; selecting a set of the hosts spanning at least one of the NUMA domains, each host in the set of hosts having a physical NUMA topology that maximizes locality for vCPU and memory resources of the VM as specified in the vCPU and memory configuration; and providing the set of hosts to a distributed resource scheduler (DRS) executing in the virtualization management server, the DRS configured to place the VM in a host selected from the set of hosts.

Abstract: The current document is directed to methods and systems that automatically deploy and manage applications that are associated with hardware dependencies. As one example, many machine-learning-based applications use specialized hardware accelerators during training phases since, in many cases, training of machine-learning-based applications and systems would be computationally intractable without the increased computational bandwidth provided by hardware accelerators. However, such hardware dependencies may prevent machine-learning-based applications from being deployed and managed effectively by widely used automated orchestration systems, and manual deployment of applications with hardware dependencies may suffer significant inefficiencies and problems related to maintenance downtime within distributed computer systems.

Abstract: In one set of embodiments, a computer system can analyze a layout of one or more storage objects of a VM to be placed within a cluster of host systems of a hyper-converged infrastructure (HCI) deployment. The computer system can further determine, based on the analyzing, that the one or more storage objects are stored on a single host system or within a single fault domain of the cluster. The computer system can then cause the VM to be placed on that single host system or within that single fault domain.

Abstract: Techniques for orchestrating and prioritizing the rebuild of storage object components in a hyper-converged infrastructure (HCI) deployment comprising a cluster of host systems are provided. In one set of embodiments, a computer system can identify a list of storage object components impacted by a maintenance event or failure of a host system in the cluster. The computer system can further determine a priority class for each storage object component in the list, where the determined priority class is based on a virtual machine (VM)-level priority class assigned to a VM to which the storage object component belongs.

Abstract: In certain embodiments, a computer system can create first and second pluralities of disk groups in a hyperconverged infrastructure (HCI) cluster, where each disk group in the first plurality has capacity storage devices of a first type and each disk group in the second plurality has capacity storage devices of a second type. The computer system can further tag each disk group in the first plurality with a first disk group tag, tag each disk group in the second plurality with a second disk group tag, and create a storage policy that includes a placement rule identifying the first disk group tag. Then, at a time of provisioning a virtual machine (VM) in the HCI cluster that is associated with the storage policy, the computer system can place the VM on one or more of the first plurality of disk groups in accordance with the placement rule identifying the first disk group tag.

Abstract: Techniques for automated datastore unavailability handling are provided. In one set of embodiments, a computer system can receive a request to bring a datastore offline and, in response to the request, identify one or more virtual machines (VMs) in a virtualized computing environment that have one or more virtual disks stored in the datastore. The computer system can then, for each of the one or more VMs, determine an action to be taken with respect to the VM in response to the datastore's unavailability and trigger execution of the action.

Abstract: The present disclosure is related to methods, systems, and machine-readable media for force provisioning using available resources. A request can be received to provision a virtual computing instance by a software defined data center according to a storage policy specified as part of the request. An insufficiency of storage policy resources to satisfy the storage policy specified for the VCI can be determined.

Abstract: A method for storage management of an object among a plurality of storage devices of a datacenter is provided. The method, in response to receiving an input on a selection item presented through a UI, determines that a manual storage management of an object is selected. The method then receives a storage policy for storing the object. Based on the storage policy, the method defines a plurality of components for the object and determines whether a set of one or more storage resources is available for storing the plurality of components. When the method determines that the set is available, for each component, the method presents the set of storage resources, receives a selection of a storage resource in the set to store the component, and updates the set based on the policy and the selection before presenting the updated set to select from for storing a next component.

Abstract: The disclosure provides an approach for high-availability admission control. Embodiments include determining a number of slots present on the cluster of hosts. Embodiments include receiving an indication of a number of host failures to tolerate. Embodiments include determining a number of slots that are assigned to existing computing instances on the cluster of hosts. Embodiments include determining an available cluster capacity based on the number of slots present on the cluster of hosts, the number of host failures to tolerate, and the number of slots that are assigned to existing computing instances on the cluster of hosts. Embodiments include determining whether to admit a given computing instance to the cluster of hosts based on the available cluster capacity.

Abstract: When a failure occurs at a host in a cluster of hosts in a virtualized computing environment, virtualized computing instances that were running on the failed host are restarted on the active host(s) in the cluster. Resources to enable the restart of the virtualized computing instances are made available by powering off virtualized computing instances that are running on the active hosts. Determination of which virtualized computing instances to power off and to power on can be performed based on power off settings and restart priority levels that are configured for the virtualized computing instances.

Abstract: Techniques for optimizing cluster-wide operations in a hyper-converged infrastructure (HCI) deployment are provided. In one set of embodiments, a computer system can receive a request to initiate a cluster-wide operation on a cluster of the HCI deployment, where the cluster includes a plurality of host systems, and where the cluster-wide operation involves a host-by-host evacuation of virtual machines (VMs) and storage components from the plurality of host systems. The computer system can further generate a set of recommendations for executing the host-by-host evacuation in a manner that minimizes the total amount of time needed to complete the cluster-wide operation. The computer system can then execute the host-by-host evacuation in accordance with the set of recommendations.

Abstract: Techniques for ensuring sufficient available storage capacity for data resynchronization or data reconstruction in a cluster of a hyper-converged infrastructure (HCI) deployment are provided. In one set of embodiments, a computer system can receive a request to provision or reconfigure an object on the cluster. The computer system can further calculate one or more storage capacity reservations for one or more host systems in the cluster, where the one or more storage capacity reservations indicate one or more amounts of local storage capacity to reserve on the one or more host systems respectively in order to ensure successful data resynchronization or data reconstruction in the case of a host system failure or maintenance event. If placement of the object on the cluster will result in a conflict with the one or more storage capacity reservations, the computer system can deny the request to provision or reconfigure the object.

Abstract: Techniques for optimizing virtual machine (VM) storage performance in a hyper-converged infrastructure (HCI) deployment comprising a stretched cluster of host systems are provided. In one set of embodiments, a computer system can identify one or more stretched VMs in the stretched cluster, where storage objects associated with the one or more stretched VMs are replicated across the sites of the cluster. The computer system can further determine, for each stretched VM, whether a greater number of the VM's storage objects are accessible to the VM via site-local replica copies residing at a first site where the VM is currently running, or via site-remote replica copies residing at a second site where the VM is not currently running. If a greater number of the VM's storage objects are accessible to the VM via the site-remote replica copies, the VM can be migrated from the first site to the second site.

Abstract: Techniques for optimizing cluster-wide operations in a hyper-converged infrastructure (HCI) deployment are provided. In one set of embodiments, a computer system can receive a request to initiate a cluster-wide operation on a cluster of the HCI deployment, where the cluster includes a plurality of host systems, and where the cluster-wide operation involves a host-by-host evacuation of virtual machines (VMs) and storage components from the plurality of host systems. The computer system can further generate a set of recommendations for executing the host-by-host evacuation in a manner that minimizes the total amount of time needed to complete the cluster-wide operation. The computer system can then execute the host-by-host evacuation in accordance with the set of recommendations.