Cisco Storage Server Node Configuration

Note

Cisco firmmware revision 4.0(4) or higher is required with Hyperscale X.

Prior to HyperScale X Reference Architecture software installation each server must meet basic minimum configuration requirements.

The following steps apply to Cisco UCS C240 and UCS S3260 server nodes only.

Cisco UCS managed and unmanaged servers will require different procedures to achieve the following configurations. Refer to Cisco documentation for specific steps required to complete this configuration.

Ensure that all steps are completed for all servers that will be part of the HyperScale X Reference Architecture Storage Pool.

Cisco UCS C240 M5 Configuration Only

1.	Boot Drive -SSD Drives	Create a RAID1 volume using the two smaller SSD or M.2 drives assigned for the Operating System(OS). Set BIOS to boot from RAID1 volume. Disable Secure Boot
2.	Metadata Drives	The two dedicated metadata 3.2TB NVMe drives require no further configuration. Commvault HyperScale ISO will detect and format drives during initial imaging process.
3.	Storage Pool Drives	Storage Pool Large Form Factor (LFF) should have no RAID and no special configuration. Commvault HyperScale ISO will detect and format drives during initial imaging process. Ensure all LFF drives are labeled as JBOD and not part of a RAID group. Continue to step 4 below.

Boot Drive -SSD Drives

Create a RAID1 volume using the two smaller SSD or M.2 drives assigned for the Operating System(OS).

Set BIOS to boot from RAID1 volume.

Disable Secure Boot

Metadata Drives

The two dedicated metadata 3.2TB NVMe drives require no further configuration.

Commvault HyperScale ISO will detect and format drives during initial imaging process.

Storage Pool Drives

Storage Pool Large Form Factor (LFF) should have no RAID and no special configuration.

Commvault HyperScale ISO will detect and format drives during initial imaging process.

Ensure all LFF drives are labeled as JBOD and not part of a RAID group.

Continue to step 4 below.

Cisco UCS S3260 M5 Configuration Only

0.	Disk Zoning	Ensure each node is assigned the correct type and number of drives. Each node should be zoned as follows: 2 rear mounted SSD drives 1 top load 3.2TB SSD drive 24 top load Large Form Factor (LFF) data drives
1.	Boot Drive - Rear mounted SSD drives	Create a RAID1 volume for the 2 rear mounted boot SSD drives to be used for OS volume. Set BIOS to boot from RAID1 volume.
2.	Metadata Drives - Top load SSD drives	Use the 6.4TB NVMe drive for the DDB/IC metadata drive, and the single top load 3.2TB SSD set to JBOD for the other metadata drive.
3.	Storage Pool Drives	Storage Pool Large Form Factor (LFF) should have no RAID and no special configuration. Commvault HyperScale ISO will detect and format drives during initial imaging process. Ensure all LFF drives are labeled as JBOD and not part of a RAID group. Continue to Step 4 below.

Both Cisco UCS C240 M5 and UCS C24 M5 Network Configuration

4.	Networking and Switch Configuration	Create a minimum of two VLAN's as follows: VLAN 1 for data protection network for public interface(s) VLAN 2 for internode network (storage-pool VLAN) for private interface(s) Ensure all server ports are connected to 10Gb (or more) ports on network switches. If multiple VLANs are required for data protection, offsite copies, etc., ensure all VLANs are configured on switches. Total number of vNIC's required will depend on the number of isolated data-protection VLAN's. If servers are NOT managed by UCS Manager, the number of server vNIC's required is 2 vNICs per VLAN. If servers are managed by UCS Manager (with fabric-Interconnects), network resilience is through fabric-failover, so only 1 server vNIC per VLAN is required.

Networking and Switch Configuration

Create a minimum of two VLAN's as follows:

VLAN 1 for data protection network for public interface(s)
VLAN 2 for internode network (storage-pool VLAN) for private interface(s)

Ensure all server ports are connected to 10Gb (or more) ports on network switches.

If multiple VLANs are required for data protection, offsite copies, etc., ensure all VLANs are configured on switches.

Total number of vNIC's required will depend on the number of isolated data-protection VLAN's.

If servers are NOT managed by UCS Manager, the number of server vNIC's required is 2 vNICs per VLAN.

If servers are managed by UCS Manager (with fabric-Interconnects), network resilience is through fabric-failover, so only 1 server vNIC per VLAN is required.