Skip to main content

Storage Pool Creation and Configuration

Screenshot 2026-03-31 at 08.38.06.png

From the system menu on the left, select the Storage Management icon to open the storage management screen. This is where the underlying storage pool is created and where access methods are configured, allowing some or all of the storage to be presented to clients. Select Go to Drive Manager to open the drive initialisation wizard.

35.png

As shown on the screen, a 1 TB drive is available and ready for initialisation. Select the spanner icon to begin preparing and formatting the disk.

Depending on the deployment, storage can be configured using either hardware RAID or software RAID. This will be determined by the available hardware and the selected file system and storage architecture.

36.png

At this stage, you will be required to assign a name to the drive. In this example, the drive is named Drive1 however, you may choose any naming convention that aligns with your environment.

Warning: This process will initialise the selected drive and all existing data will be permanently erased.

37.png

This process may take some time, particularly on larger installations, which is expected behaviour. Once complete, a confirmation message will be displayed indicating that the process has completed successfully.
Select Close to return to the drive management screen.

38.png

As shown, you are now back at the drive management screen. The drive has been initialised and is ready for use.
Under the Actions column, you can select the Identify Disk icon to illuminate the drive LED. This allows the physical disk to be identified, which can be useful when locating a specific drive, particularly in the event of a failure.

 

ZFS Pool Design and Configuration

As mentioned previously, ZFS offers a number of advanced features, but it requires careful planning and a clear understanding of your storage environment. When configured correctly, it provides a highly scalable and resilient platform, capable of growing to petabyte-scale deployments.

Screenshot 2026-03-31 at 08.38.06.png

From the system menu on the left, select the Storage Management icon to open the storage management screen. This is where the underlying storage pool is created and where access methods are configured, allowing some or all of the storage to be presented to clients. Select Go to Drive Manager to open the drive initialisation wizard. 

 

80.png

As shown, a number of 30 GB drives are available and ready for initialisation. Select the disks you want to include in the pool to begin the creation process. If you are working with a larger number of disks, you can use the Add all option at the top left of the screen to select them in one action.

 

82.png

Once the disks have been initialised, you will be presented with the option to define the ZFS RAID layout for the pool. On this screen, you define the pool layout, including the protection level and compression settings. These choices determine how data is protected, how capacity is used, and how the system behaves under different workloads.

The available protection levels map directly to RAIDZ configurations:
• Minimum – RAIDZ1 (single parity). Maximises usable capacity but provides the lowest level of protection. This should only be used for small disk groups or non-critical workloads. With larger disks, rebuild times increase significantly, raising the risk of a second failure during recovery, which can result in data loss.
• Balanced – RAIDZ2 (dual parity). Protects against two disk failures and provides a strong balance between capacity and resilience. This is the recommended option for most production workloads, including virtualisation platforms, general file services, and database-backed applications, where both availability and efficiency are important.
• Maximum – RAIDZ3 (triple parity). Protects against three disk failures and is designed for larger pools where rebuild times are extended. This is typically used in high-capacity or archive environments where maintaining data integrity is the priority over maximising usable space.
• Fast – mirrored layout. Provides the highest level of performance and the fastest rebuild times, particularly for random I/O workloads. This comes at the cost of usable capacity but is well suited to latency-sensitive applications such as databases or virtualisation environments requiring high IOPS.

Compression can be enabled to reduce the amount of physical storage consumed. ZFS applies compression inline as data is written, which can improve both capacity efficiency and, in some cases, performance by reducing the amount of data written to disk.

Compression is generally effective for:
• Virtualisation workloads (VM images, OS disks)
• File services (documents, logs, text-based data)
• Database workloads where data is compressible

It is less effective for:
• Pre-compressed data such as media files (video, images, backups)

In most cases, enabling compression provides a benefit with minimal overhead and is recommended unless there is a specific reason not to.

83.png

On this screen, you define how filesystem metadata is stored. This has a direct impact on overall system performance and should be considered alongside the selected RAIDZ layout.

You can choose to keep metadata on the same disks as the main data or place it on separate disks. Keeping metadata on the same disks is simpler, but metadata operations will share the same I/O path as regular data access, which can impact performance under load.

Using separate disks for metadata isolates these operations and can significantly improve performance, particularly in environments with mixed or random I/O workloads. This becomes more important as pool size increases or when using higher parity RAIDZ configurations, where I/O overhead is already higher.

Snapshot operations are also metadata-driven. While snapshots do not copy data when created, they rely on metadata to track changes over time. As snapshot usage increases, so does metadata activity, which can directly impact performance if not properly accounted for.

For this reason, it is recommended to offload metadata to faster storage, such as SSD or NVMe, where available. This reduces latency, improves responsiveness, and helps maintain consistent performance as the system scales.

 

Back to top