, ,

How to Add External Storage to a Workload Domain in VCF 9.1 (VCF 9.1 Day-2 Operations Step by Step Guide, Part 16)

Attach external, supplemental storage to a VCF 9.1 workload domain cluster. Step by step vSphere Client instructions for VMFS on Fibre Channel, NFS, iSCSI and vVols datastores, plus verification and fixes.

VCF 9.1 Day-2 Operations · Part 16 of 20
Task type: OPTIONAL capability. Run it only when a cluster needs shared capacity from an existing array, or when you want a separate datastore for templates, backups and data at rest.
What it does: Adds external, supplemental storage to a workload domain cluster from a Fibre Channel, NFS, iSCSI, vVols or NVMe over Fabrics array, using the vSphere Client.
Depends on: a healthy cluster in a workload domain, an array that is listed on the Broadcom compatibility guide, and host connectivity to that array already cabled or configured.
Before you begin: Confirm a healthy, fully backed up fleet before you attach any external storage. Each host should report green health, and recent backups of SDDC Manager, vCenter and NSX Manager should already exist. Work through the VCF 9.x pre-installation checklist so you start from a known good state.
TL;DR
  • External storage attaches as supplemental storage, added after a cluster exists, alongside its principal storage.
  • VCF 9.1 supports VMFS on Fibre Channel, NFS v3 and NFS v4.1, iSCSI, vVols, and NVMe over Fabrics as supplemental storage.
  • SDDC Manager does not provision or manage supplemental storage, so you do the datastore work in the vSphere Client.
  • Present the array and configure host connectivity first, then run the New Datastore wizard on the cluster.
  • iSCSI needs a software iSCSI adapter and a target address; vVols needs a registered VASA storage provider before the datastore.
  • Verify the datastore mounts on every host in the cluster and appears with green health before you place workloads on it.

This part shows you how to attach external storage to a workload domain cluster in VCF 9.1. External storage arrives as supplemental storage, which you add after a cluster is built and which runs alongside the principal storage the cluster was created with. You will confirm which protocol applies, present the array to the hosts, configure connectivity, and then create the datastore with the vSphere Client New Datastore wizard. Every datastore action runs in vCenter through the vSphere Client, since SDDC Manager does not provision or manage supplemental storage.

Treat this as an optional task. Run it when a cluster needs extra shared capacity from an array you already own, when you want a dedicated datastore for virtual machine templates or backup data, or when a workload needs a storage type that vSAN does not provide. If you want to grow vSAN capacity instead of attaching an external array, follow Part 15, expanding vSAN storage and using vSAN ESA. For the operating model behind these consoles, see Part 1 of this series.

Workload domain cluster VMFS on FC Fibre Channel array NFS v3 and v4.1 NAS filer iSCSI IP SAN vVols VASA provider Each array mounts as a supplemental datastore on the cluster SDDC Manager does not manage these datastores
A workload domain cluster can mount external arrays as supplemental datastores over several protocols. You configure and manage them in the vSphere Client, not in SDDC Manager.

Prerequisites

Confirm the items below before you attach any array. External storage must be listed on the Broadcom compatibility guide and validated as a supplemental storage solution, and host connectivity must be in place before the datastore wizard can see the device.

ItemRequirement
VCF version9.1.0.0 or later across the workload domain
Array supportListed on the Broadcom compatibility guide and validated as VCF supplemental storage
ConnectivityFC zoning, NFS export permissions, or iSCSI network reachable from every host
AccessvCenter administrator for the workload domain, plus array admin credentials
vVols onlyVendor VASA storage provider details and a storage container
HealthGreen host and cluster health, with recent SDDC Manager, vCenter and NSX backups on file

Step 1, Confirm which supplemental storage types apply

Start by matching your array to a supported supplemental storage type, since the later steps differ by protocol.

  1. Identify the array protocol you plan to use: Fibre Channel, NFS, iSCSI, vVols, or NVMe over Fabrics.
  2. Open the Broadcom compatibility guide and confirm the array and ESXi 9.0 release are listed for that protocol.
  3. Confirm the array is validated as a supplemental storage solution by the storage vendor.
  4. Record the array details you will need later, such as WWNs, NFS export paths, iSCSI target addresses, or VASA provider URLs.

Supplemental storage is any shared datastore you add to a cluster after it exists, separate from the principal storage the cluster was built on. A single cluster can run vSAN as principal storage and still mount an NFS or Fibre Channel datastore for templates, backup staging or migration. Match the protocol to what your hosts can reach today, because connectivity, not the wizard, is usually the limiting factor. The table below lists the supplemental types VCF 9.1 supports and where each one fits.

Storage typeDatastore formatTypical use
Fibre ChannelVMFS 6Block capacity from an existing FC SAN
NFS v3 and v4.1NFS datastoreFile based capacity, templates, ISO and backup data
iSCSIVMFS 6Block capacity over the IP network
vVolsvVols datastorePolicy driven array storage over FC, iSCSI or NFS
NVMe over FabricsVMFS 6Low latency block capacity over FC, TCP or RDMA

Step 2, Locate the target cluster in SDDC Manager

Use SDDC Manager to confirm the domain and cluster you will attach storage to, then move to the vSphere Client for the actual work.

  1. Open SDDC Manager and go to Inventory, then Workload Domains.
  2. Select the workload domain that owns the cluster.
  3. Open the Clusters tab and note the cluster name and its vCenter.
  4. Click the vCenter link to open the vSphere Client for that domain.

SDDC Manager owns domain level operations such as commissioning hosts and adding clusters, but it does not provision or track supplemental datastores. That boundary matters, because a datastore you add here will not appear as a managed object in SDDC Manager, and lifecycle tasks such as upgrades ignore it. Note the cluster and vCenter name so you attach storage to the correct place, then do all remaining steps in the vSphere Client. If you need to add compute capacity at the same time, see Part 12, adding ESXi hosts to a cluster, or Part 13, adding a cluster to a workload domain.

Step 3, Present the array and configure host connectivity

The wizard can only see storage the hosts can already reach, so present the array and set up connectivity on every host in the cluster first.

  1. For Fibre Channel, zone the array ports to each host HBA on the SAN fabric, then map the LUN to every host in the cluster.
  2. For NFS, export the share on the filer and grant read write access with root access to each host VMkernel IP.
  3. For iSCSI, place a VMkernel adapter on the storage network and confirm each host can reach the target address and port 3260.
  4. In the vSphere Client, select a host and open Configure, then under Storage click Storage Adapters.
  5. Click Rescan Storage so the host detects the newly presented device.

Present the array to all hosts in the cluster, not just one, or the datastore will mount unevenly and virtual machines will lose access during vMotion or failover. Fibre Channel and NVMe over Fabrics rely on correct fabric zoning and LUN masking, which you configure on the switch and array rather than in vSphere. NFS depends on export permissions and name resolution, so verify each host VMkernel IP can mount the share before you continue. A rescan after presentation is what surfaces the device to the New Datastore wizard.

Step 4, Add a VMFS datastore on Fibre Channel

Use this path for Fibre Channel and NVMe over Fabrics block storage. After the LUN is presented and rescanned, create a VMFS datastore on the cluster.

  1. In the vSphere Client, right click the cluster and select Storage, then New Datastore.
  2. On the Type screen, select VMFS and click Next.
  3. Enter a datastore Name, then select a host and choose the presented LUN from the device list.
  4. On the VMFS version screen, select VMFS 6 and click Next.
  5. Accept or adjust the partition layout, then click Next.
  6. Review the summary and click Finish to format and mount the datastore.

Creating the datastore at the cluster level mounts it on all hosts that can see the LUN, which is what you want for a shared datastore. Give the datastore a clear name that signals its role, such as templates or backup staging, so operators do not place production workloads on capacity meant for other purposes. When the same LUN is presented to every host, vSphere mounts the new VMFS volume across the cluster automatically. If a host is missing after creation, return to Step 3 and confirm its zoning and rescan.

Step 5, Add an NFS datastore

Use this path for file based capacity from a NAS filer. NFS needs no LUN, only an export path and server address that the hosts can reach.

  1. Right click the cluster and select Storage, then New Datastore.
  2. On the Type screen, select NFS and click Next.
  3. Select the NFS version, either NFS 3 or NFS 4.1, to match the filer, then click Next.
  4. Enter the datastore Name, the export Folder path, and the filer Server address.
  5. Select every host in the cluster on the host accessibility screen so the mount is consistent.
  6. Review the summary and click Finish to mount the NFS datastore.

Pick one NFS version and stay with it for a given export, because a host must not mount the same share over both NFS 3 and NFS 4.1 at once. NFS 4.1 adds multipathing and stronger security, while NFS 3 remains common on older filers, so match the version the array supports. Mounting on every host in the cluster keeps the datastore available during vMotion and host maintenance. If a later host joins the cluster, mount the same export on it as described in Step 8.

Step 6, Add iSCSI supplemental storage

iSCSI needs a software adapter and a target before you can format a datastore. Configure the adapter on each host, discover the target, then create VMFS.

  1. Select a host, open Configure, and under Storage click Storage Adapters.
  2. Click Add Software Adapter, select the software iSCSI adapter, and click OK.
  3. Select the new iSCSI adapter, open the Dynamic Discovery tab, and click Add.
  4. Enter the target IP address and port, then click OK and rescan the adapter.
  5. Repeat the adapter and discovery steps on every host in the cluster.
  6. Right click the cluster, select Storage, then New Datastore, choose VMFS, and select the iSCSI LUN to format as VMFS 6.

You can add the iSCSI target from the command line when scripting the host configuration. Bind the iSCSI adapter to the correct VMkernel port group so traffic uses the storage network rather than the management network. After discovery and a rescan, the target LUN appears in the New Datastore device list just like a Fibre Channel LUN. The command below adds a send target to a host software iSCSI adapter over SSH.

esxcli iscsi adapter discovery sendtarget add -A vmhba65 -a 10.10.20.50:3260
Choose the array protocol Present the array and map the LUN or export Configure host connectivity zoning, VMkernel, iSCSI adapter or VASA Run the New Datastore wizard on the cluster in vSphere Client Verify the mount on all hosts
The order stays the same across protocols: choose, present, connect, create, then verify. Only the connectivity step changes with the storage type.

Step 7, Add a vVols datastore

vVols gives you policy driven array storage. Register the vendor VASA storage provider first, then create the vVols datastore backed by a storage container.

  1. Select the vCenter object in the inventory, open Configure, and click Storage Providers.
  2. Click Add, enter the vendor VASA provider name, URL and credentials, then click OK.
  3. Wait for the provider to register and show an online status.
  4. Right click the cluster, select Storage, then New Datastore, and choose vVol on the Type screen.
  5. Select the storage container published by the provider, name the datastore, and select every host in the cluster.
  6. Review the summary and click Finish to create the vVols datastore.

vVols moves the unit of management from a LUN to an individual virtual machine object, so the array applies policy per virtual disk rather than per datastore. Register the VASA provider at the vCenter level once, and every cluster in that vCenter can consume containers the provider publishes. Confirm the provider shows online and healthy before you create the datastore, since a stale certificate or a blocked port will stop the container from appearing. After creation, assign a storage policy that maps to the array capabilities you want, such as replication or a performance tier.

Step 8, Mount the datastore on additional or new hosts

Keep the datastore mounted on every host so it stays available as the cluster changes. When you add a host later, mount the same storage on it.

  1. Present the LUN or export to the new host, or configure its iSCSI adapter, as in Step 3.
  2. Select the new host, open Configure, then under Storage click Storage Adapters and click Rescan Storage.
  3. For NFS, right click the host, select Storage, then New Datastore, and mount the same export path and server.
  4. For block storage, confirm the shared VMFS datastore appears under the host after the rescan.
  5. Confirm the datastore now lists the full host count under its Hosts tab.

A datastore that mounts on some hosts but not others breaks vMotion and can strand virtual machines during a host failure. When you scale a cluster with new hosts, treat the supplemental datastore as part of that work and mount it before you place workloads on the new host. For the host commission and add steps, see Part 12, adding ESXi hosts to a cluster. A quick check of the datastore Hosts tab confirms the mount count matches the cluster size.

Verify the new storage

Confirm the datastore is healthy and mounted everywhere before you place workloads on it. Check each of the following.

  • In the vSphere Client, open the cluster, then Datastores, and confirm the new datastore appears with the expected capacity.
  • Open the datastore and check the Hosts tab shows every host in the cluster as connected.
  • Confirm the datastore reports normal status with no accessibility alarms.
  • Confirm the mount on a host over SSH, then list NFS mounts if you added an NFS datastore:
esxcli storage filesystem list
esxcli storage nfs list

Each host should list the new datastore as mounted and accessible in the filesystem output. For a quick functional test, create a small test virtual machine or migrate a non production one onto the datastore and confirm it powers on and runs. Because supplemental storage sits outside SDDC Manager, record the datastore, its protocol and its purpose in your own runbook so the next operator knows why it exists. A consistent mount count plus a clean status is your signal that the attach is complete.

Common errors and fixes

SymptomLikely causeFix
LUN missing from the New Datastore device listZoning or LUN masking incomplete, or no rescanConfirm fabric zoning and LUN mapping to the host, then rescan the storage adapters.
NFS mount fails with a permission errorExport lacks the host VMkernel IP or root accessAdd each host VMkernel IP to the export with read write and root access, then retry the mount.
iSCSI target does not appear after discoveryVMkernel not on the storage network or port 3260 blockedBind the iSCSI adapter to the correct VMkernel port and open port 3260 to the target, then rescan.
Datastore mounts on some hosts onlyStorage not presented to every host in the clusterPresent the LUN or export to all hosts, rescan, and mount NFS on the missing hosts.
vVols container does not appearVASA provider offline or certificate expiredRe register the VASA storage provider, refresh its certificate, and confirm it shows online.

Common questions

Whether SDDC Manager manages external storage. No. SDDC Manager owns domain operations such as adding hosts and clusters, but supplemental datastores are created and managed in the vSphere Client. Track them in your own runbook, since VCF lifecycle tasks do not manage them.

Whether external storage can be principal storage. Fibre Channel VMFS and NFS v3 can be principal storage during greenfield cluster creation, while iSCSI, NFS v4.1 and NVMe over Fabrics are supplemental in greenfield workflows. This part covers attaching storage to an existing cluster, which is always a supplemental add.

How VCF picks the primary datastore. When a cluster has several datastore types, VCF selects the primary in a fixed priority order of vSAN, then NFS v3, then VMFS, then NFS 4.1, then iSCSI. Adding supplemental storage does not change the principal datastore the cluster was built on.

Whether you can mix NFS 3 and NFS 4.1. A single export must be mounted with one version across all hosts. You can run different exports at different versions on the same cluster, but never mount one share over both versions at once.

How this differs from expanding vSAN. Attaching an external array adds a separate datastore from a storage system you already own, while vSAN expansion grows the built in storage pool. To grow vSAN instead, follow Part 15 of this series.

Whether supplemental storage supports Storage Policy Based Management. vVols exposes array capabilities to storage policies, while plain VMFS and NFS datastores offer basic tagging rather than array driven policy. Choose vVols when you want per virtual machine policy from the array.

References

VCF 9.1 Day-2 Operations · Part 16 of 20
« Previous: Part 15  |  Complete Guide  |  Next: Part 17 »

About The Author


Discover more from Journal of Intelligent Infrastructure

Subscribe to get the latest posts sent to your email.

Leave a Reply

Your email address will not be published. Required fields are marked *

Architect’s Toolkit

About the Author

Dr. Pranay Jha is a Cloud and AI Consultant with 18+ years of experience in hybrid cloud, virtualization, and enterprise infrastructure transformation. He specializes in VMware technologies, multi-cloud strategy, and Generative AI solutions. He holds a PhD in Computer Applications with research focused on Cloud and AI, has published multiple research papers, and has been a VMware vExpert since 2016 and a VMUG Community Leader.

Discover more from Journal of Intelligent Infrastructure

Subscribe now to keep reading and get access to the full archive.

Continue reading