With your SOFS deployed, you might be wondering if you’re all set. Can you now deploy VMs and other important data onto this new, resilient storage back end? The answer, at this stage, is no, not just yet. You still need to perform several steps to aggregate the physical disks into a storage pool and slice this pool into virtual disks, which are sometimes known as storage spaces. From there, you need to transform these into Cluster Shared Volumes (CSVs), on which file shares will reside. That might seem like a significant number of steps, but fortunately, System Center Virtual Machine Manager makes it easy. The example in Figure 4-15 shows the relationship between the disks, the pools, the virtual disks, the CSVs, and, finally, the file shares themselves.
FIGURE 4-15 Layers of the Windows Server storage stack
As shown in the diagram, at the very bottom are the raw physical disks that are available through the JBODs. These physical disks are aggregated into one or more storage pools. Since this configuration uses multiple clustered hosts as part of a SOFS, these storage pools are also clustered, with each node having visibility into the pool.
From there, the total raw capacity is sliced into virtual disks, also known as storage spaces. These storage spaces can be configured to support tiering if the underlying storage pool has a mix of SSDs and HDDs available. You can also define the level of redundancy you require, with settings such as mirroring or parity.
Because this is a SOFS configuration, to ensure that these storage spaces are available across multiple nodes in the SOFS, they are transformed into CSVs. When you use CSVs with the SOFS, all cluster nodes can simultaneously write to the same storage, increasing
performance and availability.
Finally, in this procedure you create continuously available file shares that sit within the CSVs. Continuously available file shares hosted on the SOFS let you store Hyper-V VM configuration files and virtual hard disks in easy-to-manage, remotely accessible file shares without sacrificing performance or availability.
Again, although this looks like a long process, System Center Virtual Machine Manager makes it straightforward.
1. Log on to VMM01 using contoso\administrator credentials.
2. Open the System Center Virtual Machine Manager console, click Fabric, and then click Storage.
3. Click File Servers, and in the main window, right-click SOFS.contoso.com, and select Manage Pools.
4. In the Storage Pools window, click New.
5. In the Create Storage Pool window, for the name, type POOL1.
6. From the Classification drop-down list, select Primary Pool.
7. Within the Disk column, as shown in Figure 4-16, select all of the disks across both enclosures, and then click Create.
FIGURE 4-16 Disks discovered by System Center Virtual Machine Manager
As shown in Figure 4-16, the data in both the Enclosure and Slot columns is populated. If these columns are blank, ensure that you are using the latest JBOD enclosure firmware and disk firmware and that there are no disk errors.
To support deployments that require an added level of fault tolerance, storage spaces can associate data with a particular JBOD enclosure. This capability is known as enclosure awareness. With enclosure awareness, if one enclosure fails or goes offline, the data remains available in one or more alternative enclosures.
If you use enclosure awareness with storage spaces, your JBOD must support SCSI Enclosure Services (SES). With enclosure awareness in place, depending on your configuration, you can tolerate one or more failed enclosures, as shown in Table 4-2. To tolerate one failed enclosure with two-way mirrored spaces, you need three compatible storage enclosures. To tolerate two failed enclosures with three-way mirrored spaces, you need five compatible storage enclosures. This configuration has only two enclosures, so it cannot tolerate the failure of a complete enclosure.
TABLE 4-2 Storage space resiliency and SES options based on the number of JBODs
TWO JBODS THREE JBODS FOUR JBODS
2-way mirror 1 disk 1 enclosure 1 enclosure 3-way mirror 2 disks 1 enclosure + 1 disk 1 enclosure + 1 disk Dual parity 2 disks 2 disks 1 enclosure + 1 disk
As shown in Table 4-2, the configuration for this POC aligns with the Two JBODs column. When you configure Storage Spaces in future steps, the level of resilience you select will determine how many disk losses can be tolerated. As shown in Figure 4-17, POOL1 has 24 disks and is the prmary pool. The total disks include the data storage and protection.
FIGURE 4-17 Configured storage pools in System Center Virtual Machine Manager
8. On the Storage Pools page, click OK, and System Center Virtual Machine Manager creates the storage pool. In this POC configuration, and for other smaller
environments, a single pool eases administration while still providing high levels of redundancy and performance. But as the number of disks grows, it is a best practice to split those disks into separate pools to minimize the time required to fail over the storage pool to another node.
9. In the main System Center Virtual Machine Manager console window, under Storage, click Classifications And Pools. Notice that your three classifications are listed, one of which has an associated storage pool.
NOTE To view the clustered storage pool that System Center Virtual Machine Manager has created, log on to FS01, open Failover Cluster Manager, expand Storage, and click Pools. The new Storage Pool should be listed.
MORE INFO For further detailed guidance on how to optimally configure your storage pools based on your environment, review the Storage Spaces design information on TechNet at http://go.microsoft.com/fwlink/?LinkID=517497.