Home > Storage > PowerStore > Data Protection > Dell PowerStore: Clustering and High Availability > Resource balancer
In the modern data center, administrators must be quick and agile to support mission-critical applications. PowerStore offers an intelligent analytical engine that is built into the PowerStore operating system. This engine offers many benefits such as helping administrators make decisions that are based on the initial placement of data, and automatically balancing node performance within an appliance. It also assists with volume migrations or storage expansions that are based on analytics and capacity forecasting.
When storage administrators create storage resources, the resource balancer can provide many benefits. For example, when a multi-appliance cluster is deployed, the resource balancer intelligently and automatically places newly created volumes on different appliances in the cluster. This placement is based on which appliance has the most unused capacity. If you choose to override the decision of automatic placement, you can manually select the appliance on which to place the volume.
The following figure shows an example of a volume that will be placed automatically on a specific appliance. Note that when volumes or volume groups are created, they are only placed on a single appliance and do not span multiple appliances in the cluster. After these storage resources are created, they do not move until they are migrated manually to another appliance in the cluster. The migration of storage resources within a cluster is discussed in the following sections of this paper.
In PowerStore, when you create volume groups for the first time, all corresponding members in the volume group are placed on the same appliance. If you have existing volumes, there might be situations in which these volumes could have been automatically placed on different appliances in the cluster. In these situations, you must manually migrate the storage resources to a single appliance before placing them into a volume group.
After you create a volume or volume group on an appliance, capacity, and demand might change over time. PowerStore supports moving a volume, volume group, or vVol to another appliance within the cluster. You can perform this operation with a Manual Migration, Assisted Migration, or Appliance Space Evacuation in a PowerStore cluster by using PowerStore Manager, REST API, and PSTCLI. These methods are detailed in the next sections of this paper. For more information regarding vVol migrations, see the document PowerStore: Virtualization Integration.
Before the start of a migration job, the administrator must verify the following steps:
After you verify the above steps, you can start a migration job that is non-disruptive and transparent to the host. Although it is not visible to the end user, the storage resource is transferred to the new appliance within the cluster by using asynchronous replication technology. When the storage resource is being moved to the destination, all associated storage objects, such as snapshots and clones that are tied to the storage resource, are moved to the destination also. After the data transfer is complete, the host switches paths automatically to the destination appliance and the migration job is complete.
The following figure shows an example of a multi-appliance cluster that is made up of four appliances. This example involves moving a storage resource from the PowerStore 1000T model over to a PowerStore 5000T model. The host may have connectivity to the two appliances by using iSCSI, FC, NVMe/TCP, or NVMe/FC protocols. However, the PowerStore 3000T and 7000T models do not have host connectivity. This means that the storage resource would not be able to migrate over to those appliances until host connectivity is established.
PowerStore supports moving storage resources such as a volume, volume group, or vVol to another appliance within the cluster. Before moving the storage resource, ensure that paths are seen from the PowerStore Manager and that a host rescan has been performed. After the host initiator paths are verified, you can perform a manual migration on the storage resource.
Consider the following points when performing a manual migration in PowerStore for a volume, volume group, or vVol depends on the storage resource:
The example shown in the following figure depicts a volume that is named Test-Volume-001 and is created on a two-appliance cluster. In this example, the volume was created on appliance 2 and the wizard is presenting a list of available appliances in the cluster for where the volume can be migrated. When the Migrate Volume wizard appears, the other appliance in the two-appliance cluster is appliance 1, and the administrator can click Start Migration to initiate the migration job.
While the system is running, it periodically monitors storage resource utilization across all appliances within the cluster. Over time, an appliance might approach the maximum usable capacity. In this scenario, the system generates migration recommendations as shown in the following figure. These recommendations are generated based on factors such as drive wear, appliance capacity, and health. An administrator can view assisted-migration recommendations from the alerts or from the Migration > Migration Actions page in PowerStore Manager. If the administrator accepts a migration recommendation, a migration session is automatically created. Any snapshots or clones that are associated with the original storage resources are migrated over to the new appliance, as with a manual volume migration.
PowerStoreOS 2.0 introduces the ability to automatically move storage resources such as Volumes, Volume Groups, and vVols from an appliance, all within PowerStore Manager. This easy-to-use, wizard-based workflow can help you prepare an appliance to be powered off or removed from a cluster, or migrate storage resources for an appliance that is full.
To perform this task, go to the Hardware page, select an appliance, and click More Actions > Migrate. This action launches a new workflow as shown in the following figure. The wizard presents a list of storage resources that include Volumes, Volume Groups, vVols, Virtual Machines, and Replication Groups. You can optionally select all storage resources that are shown or select individual storage resources manually. All storage resources might not be available when you run the wizard. Storage resources that are in an active import session or in an active internal migration session are marked as Not Eligible for Migration. After the import sessions or migration sessions are complete, you can rerun the wizard to move the storage resources to other appliances in the wizard.
When you select storage resources, PowerStore intelligently calculates which appliance is best for the various storage resources. As shown in the following figure, PowerStore recommends that the ERP volume be moved to the destination Appliance 5200T-B. PowerStore makes these recommendations automatically, but there might be scenarios when you would like to place the storage resource on a specific appliance. As shown in the following figure, you can override the destination by clicking the pencil icon next to Appliance 5200T-B.
At the end of the wizard, PowerStore might provide a list of associated hosts that require a rescan. After the rescan is completed, PowerStore automatically generates the migration sessions and starts the migration jobs.
After migration for a storage resource is generated, an administrator can view the various sessions that are active within the cluster. Administrators can pause, resume, or delete any active migration sessions through PowerStore Manager, REST API, or PSTCLI. The following table shows a list of all the possible states for a migration session.
State | Definition |
Initializing | Migration session starts and stays in this state until the session initialization completes |
Initialized | Migration session transitions to this state when the session initialization completes |
Synchronizing | Background copy is in progress |
Idle | Migration session transitions to this state when the initial background copy completes |
Cutting_Over | A final portion of the copy is performed in this state, and the ownership of the storage resource is transferred to the new appliance. |
Deleting | Migration session is being deleted |
Completed | Migration session is completed, and it is safe to delete the session |
Pausing | Migration session transitions to this state when the pause command is issued |
Paused | Migration session is Paused, and user intervention is required to resume the session |
System_Paused | Migration session transitions to this state if it encounters any error, and user may resume or delete the migration after resolving the error |
Resuming | Migration session background copy is resumed |
Failed | Migration session encountered an error |
PowerStore provides details on when a cluster might run out of space. These details help when planning to scale a cluster based on business needs. To inform on these decisions, PowerStore maintains historical statistics about the consumed storage on an appliance to intelligently predict how this storage will be used over time.
To provide the most accurate information, the system collects statistics over 15 days. After 15 days, PowerStore Manager displays a forecast of when a system might run out of space. The following figure shows an example of viewing capacity forecasting in PowerStore Manager.
When mapping volumes to hosts, PowerStore dynamically sets a node for preferred host access, also known as Node Affinity, as shown in the following figure. As more storage resources are created and used over time, there might be situations where one PowerStore node is more heavily used than the peer node. In this scenario, PowerStore might automatically change the Node Affinity of one or more storage resources if certain performance metrics checks are met.
The first performance check is to compare the read and write IOPs of the source and destination nodes. The second performance check is to ensure that at least one of the nodes has greater than 50% CPU utilization. The last performance check is to see if the system is exceeding a latency check as reported by the data path engine. PowerStore checks these performance metrics every 30 minutes. If all three of these performance checks are met throughout the time duration of 30 minutes, PowerStore dynamically changes the node affinity of the volume. This change is transparent and non-disruptive to the host. PowerStoreOS 4.0 and later expands dynamic node affinity to support vVols as well. It leverages the same resource balancer engine that is used for volumes and provides the same functionality and benefits.