PowerVM Shared Storage Pool (SSP) is a pool of common storage that can be shared across multiple Virtual IO Servers (VIOS) and can be part of single or multiple Power Servers. All the Virtual IO Servers that are part of SSP are called Nodes in the Cluster. SSP enables virtualization of a pool of storage where the storage virtualization is managed at the server rather than at the back-end storage controller. This enables tighter coupling of storage virtualization with server virtualization, simplifying and optimizing storage management in virtualized environments. The close integration of storage virtualization with server virtualization also enables opportunities for more optimized use of storage and advanced storage virtualization capabilities.
Advantages of PowerVM Storage Pool :
- Optimized use of storage
- Storage on the go (Over commitment of Storage)
- Redundancy of the storage
- Provides storage based on performance requirements
Power Hardware Management Console (HMC) provides a way to create and manage PowerVM Shared Storage Pool Cluster (SSP Cluster). Below are the SSP operations that can be performed using HMC:
- SSP Cluster creation and management.
- Multiple Tier support and its management (Tier - Logical grouping of different Physical Volumes)
- Pool Mirroring (Redundancy support)
- Logical Unit life cycle management (Create, Assign, Increase size, Migrate to different Tier)
1. View Shared Storage Pool Clusters:
You can view all the Clusters that are being managed by the HMC by navigating to the "All Shared Storage Pool Clusters" Option in the Resources Menu.
1.1 Creation of Shared Storage Pool Cluster:
HMC will support the creation and management of SSP Cluster V4 and above. Below are the prerequisites for SSP Cluster creation.
1.1.1 Prerequisites to create SSP Cluster:
- HMC V8R810.1 or later (Single Tier) / V8R840.0 or later (Multi Tier and Pool Mirroring support)
- Virtual IO Server with active RMC: Version 220.127.116.11 or later (Single Tier) and Version 2.2.4 (Multi Tier and Pool Mirroring support)
- Hostname has to be configured for Virtual IO Server
- Common storage across Nodes that are backed by Fibre Channel Adapter
- Two Physical Volumes, with minimum of one 1GB and the other with 10GB capacity
To create SSP Cluster click on the Add Shared Storage Pool Cluster option as shown in Fig 1.1, this will launch a SSP Cluster creation wizard as shown the Fig 1.2.
HMC will pre-populate with certain default values for all the required attributes but the user can always modify the default values.
- You can choose to create a SSP Cluster with Single Tier or Multiple Tier capability. If you choose Multiple Tier Capability, only Nodes (2.2.4 and later) which are capable of Multi Tier will be listed in the subsequent panel.
- The threshold values for available free space and over-commit are set at Pool level during SSP Cluster creation. However these values will be obsolete when multi-tiers are created and threshold have to be set at tier level only.
To select the nodes that are to be part of Cluster creation, click on the "Next>" button, the wizard will list all the suitable Nodes based on the Tier capability selected. On selecting a particular VIOS to be part of the cluster, HMC will dynamically filter the node list and will list the Nodes which share common storage.
You can create SSP Cluster with a combination of both single tier and multiple tier capable Nodes, but even if any one of the Nodes selected is non-multi tier capable then SSP Cluster will not support multi tier. To make your SSP Cluster multi tier capable you need to remove all the non-multi tier capable Nodes from the SSP Cluster.
Once Nodes are selected, on the click of the "Next" button, the wizard will allow you to select Physical Volume (with minimum of 1GB capacity) that would be used as Repository disk. This disk will be used for bookkeeping of SSP Cluster.
While creating a cluster, only one Tier can be configured. You can add additional tiers later, if required. Mirroring can be enabled for the tier being created.
In the next panel, you can select the Physical Volumes that would be part of Tier or the Failure Groups. When mirroring is enabled SSP will be created with two Failure Group and both failure groups will be a replica of each other to support redundancy. Hence you need to assign Physical Volumes to both failure groups.
If size of both Failure Groups are not the same then the effective capacity of Tier / Shared Storage Pool will be of capacity of failure group with the least size. Hence you can see a warning/information icon in Fig 1.5. The extra capacity allocated to the other failure group will be wasted, so it's always a good practice to make sure Physical Volumes are selected such that both Failure Groups are of the same size.
You can review the configuration of SSP Cluster being created by the click of the "Next" button. The wizard will display a summary of the Shared Storage Pool Cluster. If you wish to change any of the configuration, you can go back and change before clicking on the "Finish" button.
Once you review the configuration of SSP Cluster click the "Finish" button to create SSP Cluster. It might take a while to create the SSP Cluster; after successful creation you can view all the Clusters that can be managed by the HMC as shown in Fig 1.7.
2 Shared Storage Pool Cluster Management:
You can manage or work with a specific SSP by clicking on the SSP, which will launch a new panel with the manage options.
HMC provides an option to add Tier to the SSP Cluster. On the click of "Add Tier", a new panel is displayed to create a Tier with user defined name, threshold values and allows to enable or disable pool mirroring on the Tier that will be created as shown in Fig 2.2
2.1 Tier Operations:
HMC provides a lot of Tier management options. You can select a Tier to work on and click the "Action" button which will display the drop down menu with Tier operations:
Below are the list of operations that can be performed on a Tier :
- Rename Tier: You can modify the user defined name of a Tier using this option.
- Modify Threshold: You can modify the free space and overcommit threshold values for the tier.
- Set as Default: If a tier is marked as default, then any of the Logical Units that are created without specifying Tier will be created on the default Tier.
- Restrict Tier: This operation can be performed only on System Tier. Once the System Tier is marked as Restricted, Logical Units cannot be created on it, which means System Tier will only be used for book keeping and not for data storage. Logical Units that are already existing will continue to exist unless the user migrates it to another Tier.
- Un-Restrict Tier: This operation is allowed only on Restricted System Tier. Once System Tier is marked as Un-Restricted, users can create Logical Units on System Tier (Not a recommended configuration).
- Remove Tier: This operation allows removal of all user tiers. The System Tier cannot be removed. Deleting SSP Cluster will implicitly delete the System Tier in it.
- Enable Mirroring: You can enable mirroring on a Tier if it is not mirrored; this allows creation of a new Failure Group of same capacity or greater than the existing capacity of Tier.
- Disable Mirroring: If the Tier is already mirrored then HMC provides a way to remove the Failure Group which disables mirroring and the data on the disk are no longer redundant.
- Add Capacity: This operation will allow you to increase the capacity of the Tier by adding more Physical Volumes to Failure Group.
- Remove Capacity: This operation will allow you to decrease the capacity of Tier by removing the Physical Volumes from Failure Group. Since data might reside on Physical Volumes being removed, this operation might take a while to complete.
- Rename FG: HMC provides the option to modify the user defined name of failure group. On click of "Rename FG" a panel will launch and you can specify a new name to the existing failure group (as shown in Fig 2.4).
- Replace Disk: You can replace any specific disk with another disk or set of disks whose total capacity is the same or greater than the disk being replaced. Since data might reside on Physical Volumes being removed, this operation might take a while to complete (as shown in Fig 2.4).
2.1.2 Best Practices for Tier Management:
1. It is always recommended to have at least 2 Tiers in the Pool (one System Tier and one User Tier). 2. It is recommended to set a User Tier as default and mark the System Tier as Restricted.3. If the data on the Tier is critical then it is recommended to enable mirroring. Once mirroring is enabled all the Logical Units created will have copies on both the Failure Groups.
- If mirroring is enabled effective capacity of Tier will only increase if Physical Volumes are added to both of the Failure Groups.
- If mirroring is enabled and user tries to remove the only or the last Physical Volume of a Failure Group then the Physical Volume is allowed to be removed by disabling pool mirroring on the Tier.
- Remove Tier is possible only if it does not contain any SSP Volumes. If SSP Volume exists and is mapped to any partition, then user can either migrate SSP Volume to another Tier or un-assign SSP Volume from Logical partition. Thereafter remove SSP Volume and try deleting Tier.
3. Assign Shared Storage Pool Volumes to Logical Partition:
The Virtual Storage view of Logical Partition in the HMC provides option for users to create SSP Volumes (Logical Units) and assign it to the Partitions.You can click on Add Virtual SCSI Storage, which will launch a panel to select the type of backing device for the virtual SCSI storage. Choose Shared Storage Pool Volumes as backing device and then select Add new Shared Storage Pool Volume as shown in Fig 3.2.
If thin provisioning is selected while creating SSP Volume then the size with which it is created is only allocated but not reserved, which helps SSP to overcommit the storage. The SSP Volumes can gradually grow up to the allocated size.
If thick provisioning is selected while creating SSP Volume then the size with which it is created is not only allocated but also reserved which means storage is guaranteed any time.
4 Shared Storage Pool Operations:
HMC has two launch points to manage SSP Volumes:
- From the Virtual Storage panel of the selected Logical Partition
- From Resources --> All Shared Storage Pool Cluster --> Cluster --> Tier and click on the Shared Storage Pool Volumes Tab
The supported operations on SSP Volumes are:
- Increase Size: You can increase the allocated size of a SSP Volume which helps SSP Volume to grow on a need basis.
- Migrate to Different Tier: You can migrate SSP Volumes from an existing Tier to any other Tier on the same SSP Pool.
- Remove : You can remove the SSP Volumes that are not being used by Partitions any more.
- When migrating SSP Volumes to different Tier, make sure there is sufficient free space on the destination Tier which is same or greater than the capacity of the Shared Storage Pool Volume.
- When removing SSP Volume make sure the volume is not assigned to any partition.
5 Node Operations :
HMC provides option to add or remove Nodes from the SSP Cluster. As shown in Fig 5.1 you can click on "Add Node" to add more Nodes or select the Node already existing in the SSP Cluster to remove.
Add Node option will only list Nodes which share common storage that are being currently used by the SSP Cluster.
6 Remove Shared Storage Pool Cluster:
HMC allows the removal of the SSP Cluster. The launch point for this operation is from "All Shared Storage Pool Cluster" page. You can select the SSP Cluster to be removed and click on "Action" button which provides drop down menu to Remove Cluster.
SSP Cluster can only be removed if it does not consists of any SSP Volumes for any of the Tier that are part of SSP Cluster being removed.
You can also create/deploy a partition with an SSP Volume and capture a partition configuration using SSP Volumes as a template which we will explain more on a future post.
You can alsp programmactically create and manage SSP Cluster using HMC REST API services.
Contacting the PowerVM Team:
Have questions for the PowerVM team or want to learn more? Follow our discussion group on LinkedIn IBM PowerVM