The following PowerOne components are assembled to provide an integrated system architecture:
Each component in PowerOne has a component state that describes its operational status. The component state determines the actions that are available for the components.
After a workflow that configures a particular component is completed in PowerOne, the state of that component is updated automatically. PowerOne also updates component states during regular inventory scans.
In PowerOne Navigator, the system detects any component in the New Asset state, and allows you to apply a baseline configuration to the component that changes its component state to Automatable.
In PowerOne Navigator you can change the compute sled's firmware if the component is Automatable and if that compute sled is not part of a cluster.
The PowerOne Controller performs several operations that manage PowerOne. It provides a dedicated unified management plane for PowerOne compute domains, PowerOne Pods, PowerOne networks, and PowerOne storage domains. It orchestrates these components together into a unified group of PowerOne components known as a Cluster Resource Group (CRG).
The PowerOne Controller is a dedicated system reserved solely for preinstalled PowerOne management workloads. It is a highly-available platform that hosts the PowerOne Controller software. It consists of three PowerOne Controller rack servers that are clustered to create a single unified platform for using the PowerOne Controller software. Each of the three PowerOne Controller servers contains flash storage.
The PowerOne Controller nodes are directly accessed only by the PowerOne API or PowerOne Navigator.
The following figure shows the PowerOne Controller and the PowerOne management network infrastructure.
Dell EMC NetWorker provides data protection for the PowerOne Controller. The data protection system uses two separate virtual machines (VMs) to protect the PowerOne Controller. One VM is deployed in the PowerOne Controller (called Bastion), the other VM is a NetWorker VM (called NetWorker Server) deployed on the vSphere Management Cluster.
Bastion is responsible for collecting backups of the PowerOne Controller and transmitting them to the NetWorker Server. The NetWorker Server has a 1 TB VMDK virtual disk (Eager Thick provisioned) on a VMFS volume in the vSphere Management Cluster. The VMFS volume is provisioned on the PowerOne storage domain. This configuration creates separate production and backup volumes, which allows the PowerOne controller to recover in case of failure.
NetWorker performs full-image level backups of the PowerOne Controller daily.
Backups of the PowerOne Controller are retained for three days, assuming you perform a single backup each day. If you require longer retention periods or more frequent backups, you can increase the NetWorker license from 1 TB to the appropriate size for your data retention needs.
Optionally, you can purchase Data Domain to expand your data protection capacity for the PowerOne Controller. If you select Data Domain to protect production workloads in PowerOne, you can use the same Data Domain to protect the PowerOne Controller. This reduces the number of data protection devices in the environment.
The PowerOne Pod is a collection of components, physically separated into compute domains that connect northbound to a single pair of S5232F-ON switches.
A PowerOne Pod consists of the following:
The following figure shows a PowerOne Pod consisting of three compute domains, each having two S5232-ON edge switches and two OOB management S4148T-ON switches. It also shows the interconnections between the MX7000 chassis.
One of the core components of a CRG is its computing capacity. Computing capacity is distributed among logical blocks known as PowerOne compute domains. A compute domain is a collection of interconnected Dell EMC PowerEdge MX7000 chassis in PowerOne. Each compute domain consists of a minimum of two and a maximum of ten MX7000 chassis. The first two chassis in a compute domain are designated as primary chassis and have one pair of Dell EMC MX9116n Fabric Switching Engines that provide PowerOne System Fabric connectivity. Each of the remaining eight MX7000 chassis connects to the primary MX7000 chassis using the Dell EMC MX7116n Fabric Expander Modules.
Each PowerOne Pod contains a pair of OOB management S4148T-ON switches and two S5232-ON Leaf switches. These OOB and Leaf switches form the boundaries of a PowerOne Pod. The example compute domain in the following figure shows:
The compute domain in the example provides a total capacity of 80 two-CPU (MX740c) compute sleds.
A minimum of one compute domain is required within PowerOne, and each PowerOne Pod must contain at least one compute domain. PowerOne Pod is a collection of one to three compute domains, and associated management and network switches.
Compute domains scale at the MX7000 chassis level, starting at a minimum of two MX7000 chassis to a maximum of ten chassis per compute domain. The MX7000 chassis scale at the compute sled level and can hold up to eight MX740c compute sleds or four MX840c compute sleds.
Only two MX7000 chassis are required for each compute domain. The first compute domain in PowerOne must be populated with a minimum of four compute sleds in each of the first two chassis, for a total of eight compute sleds across the two chassis. Four compute sleds are used for the CRG that hosts the vSphere Management Cluster, which contains management applications such as VMware vCenter. The second CRG created consists of four or more compute sleds and represents the workload cluster in a VMware Validated Design (VVD).
The storage domain provides production workload storage from managed arrays to one or more compute domains. The storage domain is a collection of supported storage components of PowerOne that are automated by the PowerOne Controller.
The storage domain supports an all-flash storage and consists of a minimum of one PowerMax array.
When you create a CRG, PowerOne allocates the appropriate storage (capacity and SAN paths) from the available storage resources. PowerOne then maps that storage volume to the compute hosts in the CRG. This process is repeated when additional storage is added to the CRG.
The following figure shows two volumes from the storage domain that are mapped to CRG (y). This is important because the storage and compute resources are being mapped together at the physical level. This mapping occurs before the CRG resources are managed by a hypervisor or an operating system. A unique function of PowerOne is to map the components to a CRG before the deployment of a hypervisor.
This grouping of components at the infrastructure level enables consumption of disparate resources as a single logical unit, a CRG. Thus, even before a hypervisor is applied, a logical mapping is performed on the various components consumed. For example, as a CRG is expanded by adding more compute capacity, it is not necessary to manually map existing volumes to the newly added compute resources; PowerOne performs this task.