Nimble SFA Volume creation and presentation to Veeam
Following on from Part 2 – Initial array setup , of this blog series, we will take a look at how to create a new volume on the array and how to present it Veeam for use as a backup repository. There are a few considerations to make during the setup which I will cover here.
The SFA is block based storage. It does not run any file services like other traditional deduplcation appliances so with this in mind, the storage needs to be presented to the host using either a Fibre Channel HBA or iSCSI software initiator / HBA. I will be using the software initiator on a Windows server-based system as Nimble has a very nice iSCSI management interface that is part of the Nimble Windows Toolkit. The Nimble Windows toolkit also checks for any issues with Windows patches during installation as well as altering iSCSI timeout values etc. If a critical patch is not installed that may have an impact on iSCSI traffic, it will alert you to this fact.
Create a volume on the SFA
To create a volume, log into the array, click on Manage / Data Storage and click the + sign to start the new volume wizard.
Define the settings as required for the volume. Name, Size, Access policies and Performance policy.
Lets take a closer look at the Backup Repository performance policy. Notice that the policy has Deduplciation enabled by default. The other policies available do not. Makes sense.
You may have noticed I have access policies set as well. Access will be limited by IQN and Chap access. These can be setup in the locations identified below.
Add the volume to a Windows server
As I mentioned earlier, the storage needs to be added via iSCSI connection and I will be utilising the Nimble Connection Manager that is part of the Nimble Windows Toolkit to facilitate this.
Launch Nimble Connection Manager and click Add. it should discover your SFA array. I will assume you know which NICs to include and exclude for MPIO access.
Once you have added the array, switch to the Nimble Volumes tab. The volume you just created should be listed.
Highlight and click connect. Define CHAP credentials if you have configured CHAP access.
Now switch to Windows disk manager. You should see a new uninitialised disk. NOTE If you have not done this already, run a command prompt as administrator and run the command “Diskpart Automount Disable” . This will stop volumes from auto mounting and potentially asking to format the partition. Initialise the disk.
Format the volume and assign a drive letter. NOTE depending on how large the volume will eventually end up, select an appropriate allocation unit size. 4KB has a maximum volume size of 16TB. Check out this Microsoft post for more info on NTFS allocation unit sizes.
Add the storage to Veeam as a backup repository
Launch Veeam backup and Replication, click into backup infrastructure and then select add backup repository.
Give the repository a name.
Choose storage type Microsoft Windows Server. Do not choose Deduplicating storage appliance.
Choose the volume you created earlier in disk manager
Note the following options as they deviate from out the box settings. Disable the limit maximum concurrent tasks. There is some debate on whether the limit concurrent tasks should be enabled, especially is utilising ReFS zero block clone technology in Windows server 2016. For the purposes of this setup though, I am sticking with the recommendations from Nimble. Click on advanced.
Set the advanced options as follows.
The Veeam integration guide states the following as to why the options above should be set:
- Align backup data file blocks. The SFA uses a fixed block size for deduplication. Veeam aligns the VM
data that is saved to a backup file to a 4 Kb block boundary. This option provides better deduplication
across backup files, but it can result in greater amount of unused space on the storage device and a higher
level of fragmentation.
- Decompress backup data blocks before restoring. This option allows any data that is compressed across
the network during a backup to be uncompressed before it lands on the SFA. Decompressing compressed
data enables better deduplication rates on the SFA.
- Use per-VM backup files. This option allows each VM being backed up to the SFA to be in its own backup
chain. By default, Veeam places all of the VMs in a job into a single backup chain. For example, if the
backup job contains 10 VMs, those 10 VMs are placed into one large backup chain. Per-VM backup would
place each VM in this example into its own VM backup chain. The SFA can achieve better performance
from this setting because it allows each VM to be its own data stream. Creating multiple streams of backup
data produces better performance. This setting increases the write queue depth on the SFA.
Clicking OK and Next pops up with the following useful titbit. Veeam 9.5 loves to push ReFS for the zero block clone integration. The reason it popped up is I have the NTFS volume mounted on a Windows Server 2016 server. Server 2016 supports zero block clone technology but the NTFS volume does not, it would need to be formatted with ReFS for this to work, hence the popup. For the purposes of my testing though, I did not want to muddy the waters for the deduplication levels by introducing another space-saving technology.
Define the mount server for vPower NFS.
Review the settings.
And you are done. The new repository will appear in the list of available repositories.
Check out PART 4 for Veeam job configuration.
- PART 1 – What is the Nimble SFA?
- PART 2 – Initial setup of the SFA.
- PART 4 – Creating a Veeam backup job and what to look out for.
- PART 5 – A closer look at the deduplication.
- PART 6 – Putting your backup data to work.