How to Connect Devices
Parameter patterns, read-only roles, and prerequisites that apply across all collectors.
Every Visual One collector follows the same shape: pick a device type, supply connection details, and supply a read-only credential. The differences between platforms are in the detail, and those details are what this page explains. For the exact values for a specific platform, use the Supported Devices reference.
The parameter pattern
Most collectors ask for three things: an address, a user, and a password. Beyond that, several platforms need additional parameters, and a few reuse the standard fields for something else entirely.
| Parameter | What it usually holds | Notable exceptions |
|---|---|---|
| IP | IP address or DNS name of the device or its management endpoint | Cohesity Helios expects a Helios Cluster ID rather than an address. AWS expects aws.amazon.com. Azure expects the Tenant or Directory ID. Rubrik Security Cloud expects <IP>.my.rubrik.com. NetApp 7-Mode accepts a comma separated list, one entry per node. |
| User | User name on the device | Azure expects a Client ID, AWS an Access Key ID, and Hitachi VSP 360 an OAuth2 Client ID. VMware needs vSphere login format, typically username@domain. Several Cohesity collectors do not use this field at all. |
| Password | Password for that user | Often an API key or secret instead: Cohesity Helios, HNAS over API, Azure Client Secret, AWS Secret Access Key, and Hitachi VSP 360 OAuth2 Client Secret all pass their key here. |
| Port | Override when not using the default | Required rather than optional for 3PAR, Spectrum Protect, and DB2. |
| Serial Number | Identifies one array where a management endpoint fronts several | Required for Compellent, VMAX/PowerMax, Hitachi Ops Center, and Hitachi VSP 360. |
| CLI Path | Local path to a vendor executable | Required for VNX (navisseccli.exe) and Spectrum Protect (dsmadmc.exe). |
| API Token | Token instead of, or in addition to, a password | Required for FlashBlade and NetBackup. Optional but needed for file data on Pure FlashArray. |
Roles are read-only by design
Collectors are built to run with the lowest privilege that still returns complete data, and the required role is different on every platform: Viewer on Cohesity, Browse on 3PAR, Monitor on IBM FlashSystem, readonly on NetApp and Pure, Prism Viewer on Nutanix, Observer on Qumulo and StoreOnce.
A few platforms genuinely need more, and it is worth knowing which before you plan an install:
- Windows hosts and Hyper-V require administrator permissions, because collection runs over WMI.
- Dell EMC VNX can report utilization and configuration with an Operator account, but performance data requires root.
- NetApp 7-Mode requires an administrator role for CLI access.
- Hitachi HNAS needs
supervisorover CLI for detailed volume data, though the preferred API collector uses a scoped read-only API key instead. - Oracle Exadata needs Cell Monitor across the compute node and every storage cell, with host equivalence configured.
Primary and legacy collectors
Several platforms expose more than one device type, because a newer API collector has superseded an older CLI one. Choose the primary type for anything new. The legacy types remain available for older firmware and for environments where the newer path is blocked.
| Platform | Primary | Legacy or alternate |
|---|---|---|
| IBM FlashSystem, Storwize, SVC | IBM_FLASH_SYSTEM | Eight per-model SSH/CLI types that share one collector |
| NetApp C-Mode | NETAPP_CLUSTER_API | NETAPP_CLUSTER |
| Nutanix | NUTANIX_V4_DEVICE | NUTANIX_DEVICE |
| Hitachi HNAS | HITACHI_HNAS_API_DEVICE | HITACHI_HNAS_DEVICE |
| HPE StoreOnce | HP_STOREONCE_API_DEVICE | HP_STOREONCE_DEVICE |
| IBM HMC | HMC_DEVICE | HMC_CLI_DEVICE |
| Cisco switches | CISCO_API | CISCO_SWITCH |
| Rubrik | RUBRIK_RSC_DEVICE | RUBRIK_BACKUP |
Note: For the IBM legacy types the choice is cosmetic. They all share one underlying SSH/CLI collector, so the type you pick only changes how the device is labelled in Visual One.
Collectors that need software installed locally
These collectors call a vendor tool or SDK on the data collection machine, so the software has to be present before the collector will run at all.
| Device | Prerequisite on the collection machine |
|---|---|
| Cohesity Backup | Cohesity Management SDK: pip install cohesity-management-sdk |
| Dell EMC VNX | EMC Navisphere CLI, with the path to navisseccli.exe supplied as CLI Path |
| IBM DS8000 | IBM DSCLI, with the directory containing dscli.exe on the system PATH |
| IBM Spectrum Protect / TSM | TSM Administrative Tools, with the path to dsmadmc.exe supplied as CLI Path |
| NetApp E-Series | SANtricity Storage Manager, with smcli.exe on the PATH |
| MSSQL (ODBC) | ODBC Driver 18 for SQL Server |
| DB2 | pip install ibm-db --only-binary=:all: and pip install ibm-db-sa --only-binary=:all: |
| GCP | The service account JSON key file, downloaded locally |
| Linux hosts | sysstat installed and enabled on the target host, not the collector |
One collection, two roles
A Nutanix collection reports the cluster both as a storage device and as its hypervisor, ESXi or AHV, in a single pass. If a vCenter runs entirely on Nutanix, you do not need a separate VMware collection. If a vCenter spans Nutanix and other hardware, talk to Visual One first, because the overlap can double count.
vSAN behaves similarly in the other direction: it collects through vCenter but reports the vSAN as a storage device alongside traditional arrays.
Cloud accounts need billing exports
Cloud collectors read cost data from the provider’s own billing export, which has to exist before cost reporting will populate.
- AWS: a Cost and Usage Report configured at a daily interval with ZIP compression.
- Azure: an App Registration granted Reader and Cost Management Reader on every subscription you want included.
- GCP: a Cloud Billing export with detailed usage cost data, in a BigQuery dataset whose name matches the project name, plus the Cloud Resource Manager, Cloud Billing, Cloud Monitoring, and Compute Engine APIs enabled.
Related pages
Supported Devices for per-device detail, VSIClient Administration for agent operations, and Troubleshooting for diagnosing a collection that does not arrive.