Itzikr's Blog

So far, we have covered the following aspects of Dell Technologies PowerStore:

High Level Overview,   Hardware,   AppsON,  vVols,  File Capabilities,  User Interface ,  Importing external storage, PowerStore local protection works , remote replication . & VMware SRM integration

If you are working for a medium/large IT organization, one of your basic and routine tasks as storage admin arriving to the office (or logging to company VPN during these covid19 quarantine days) , is managing the resources utilization: Trying to maintain the resources properly aligned with company infrastructure capabilities, preventing critical events like out-off space or bottlenecks.

Nothing too complex, just managing tens, hundreds, thousands of applications with their own utilization, space, protection requirements, data reduction patterns…

If besides your storage-admin roles you are also managing VMWare infrastructures, you are definitely familiar with VMware DRS functionality. DRS is one of the “flagship” features of VMWare. For almost 15 years DRS has been balancing loads across ESX clusters with DRS user adoption reaching more than 80%.

Until recently, this inherent load balancing concept wasn’t adopted in the storage industry. One of the reasons is that the data relationships between storage objects are complex. Balancing storage loads requires dealing with a “family” of local & remote protection relationships, data reduction considerations and the requirement to support non-disruptive application data flow.

In PowerStore we are happy to introduce a revolutionary solution that provides “DRS-like” behavior to a storage platform. PowerStore “Resource Balancer” functionality that will allow users to:

1. Track the appliance utilization, using Artificial Intelligence (AI) algorithms to create future utilization forecasts
2. As part of new volume creation and based on utilization forecasts predictions, detect and recommend the most appropriate appliance in the storage cluster to assign the resource.
3. Continuously track and make migration recommendations to balance the existing resources across appliances based on forecasts predictions that are updated every 5 mins.
4. Migrate the storage resources between appliances non-disruptively following user approval.
5. Manage the entire process supporting existing data reduction policies and data protection considerations.

Sounds fantastic ? So, let’s take a look how it’s works?

PowerStore resource balancer uses analytics to balance storage resources (volumes or volume groups) across appliances within a cluster. Resource balancer assigns resources to an appliance based on the current storage trends and forecast. Once a volume is placed on an appliance, the resource balancer proactively monitors the placement decision and will auto-generate migration recommendations from one appliance to another, if needed.

The resource balancer automatically determines the placement of new resource unless changed by the user. Resource balancer assigns resources to an appliance based on the current storage trends and forecast. You may optionally assign a volume to an appliance manually.

Resource balancer looks at system limits to determine the number of volumes and volume groups and places/migrates the resource accordingly.

Resource balancer also considers the following when making a volume placement/migration decision:

• Whether the appliance is offline
• Appliance failure status
• Capacity utilization
• Appliance connectivity to Host mappings associated to an existing Volume

The following displays the List Volume view. The Appliance column is not shown by default, click the Show/Hide Table Columns icon and check the box for Appliance if you wish to see on which Appliance the volume was created.

Volume creation and placement

When a user creates a new volume, the system will by default auto-place the volume. However, the user can override the system recommendation, and set the appliance to create the new Volumes. The list of available appliances is displayed when the user opens the Placement drop down list menu.

Volume migration recommendations

Capacity Forecasts

Capacity monitoring and forecasting require at least a 15-day runtime in order to fully predict capacity statistics. Runtime starts when the appliance comes online after initial configuration. On the 16th day, you should see forecast metrics being reported in PowerStore Manager or REST API.

The example displays the Physical Capacity from the Dashboard > Capacity Tab. In the example on the left, the system has not yet reached the required 15 days of runtime. In this case, a message is generated to inform the user that insufficient data has been collected. In the middle example, you have at least 25 days until full. In the right example, the capacity of the appliance is at 96% used and has an estimate of 8 days before it is full.

Capacity forecast display

The forecasted capacity is displayed in the Dashboard Capacity tab.

“Out of space” user alerts are displayed in the PowerStore Manager Dashboard. The Alert displays the time forecasted Time to Full in days.

When an alert is generated, go to Monitoring > Alerts page for details.

A Major alert is generated when:

• The forecast projection to run out of space is at 14 days or less.

A Minor alert is generated when:

• The forecast projection to run out of space is 15 to 28 days.

It is important to note that capacity monitoring and forecasting focuses on when an appliance could run out of capacity, not how full the appliance is. PowerStore can forecast up to 1 year with a 2-year retention.

The following example shows two Alerts generated stating the appliance is nearing full capacity. Select the Alert for suggestions on remediation.

From the Alert page, we see the Repair Flow indicates an Assisted Migration. Click on the text to view additional details and recommendations for remediation.

Other options include Cleaning Up System and Add More Drives. Selecting one of those options links the user to the General Support page for remediation.

There are two methods to migrate storage resources to other appliances in the cluster. You can manually migrate storage resources, or the system can assist in the resource movement. The Assisted Migration process follows the same steps as a manual migration.

When “Assisted Migration” link is selected from the Alerts pane, the Resource Balancer recommendations are presented with all the relevant details supporting user decision to continue with data migration:

When the user accepts the Resource Balancer recommendation and clicks on the “Migrate” button, the “Rescan Host” warning will notify that SCSI rescan is required on host side to support non-disruptive operation from Host OS perspective:

When the Rescan is confirmed click on the “Start Migration”. The non-disruptive data migration will begin for all the involved objects. The user can track and manage the process on the “Migration” screen:

When the migration is finished and the status of migration changed to “Completed”, the volume, including his “Snapshots family” is has fully migrated to the target appliance:

Below, you can see a demo of the Resource Balancer feature