Plan storage before placing an order

Plan storage before provisioning a dedicated server

The data center is in Tokyo, Japan. Decide how to balance capacity and redundancy, then configure the server, arrange payment, review delivery and connect for the first time.

Use the configuration page to place an order and the control panel to check its status and delivery details. No specific delivery timeframe is promised on this page.

  1. 01Choose a configuration
  2. 02Confirm payment
  3. 03Review delivery
  4. 04Connect for the first time
01 · Configuration

Decide how to use each drive before choosing options

A backup target usually calls for maximizing usable capacity, while a volume for frequently accessed data may need redundancy. Drive capacity alone does not tell the whole story: mirroring reduces usable capacity, and using two drives independently does not protect against drive failure. Before ordering, identify the data and backup volumes and the recovery source, then choose a configuration.

Prioritize capacity: use drives independently

The standard configuration has 2 × 1TB NVMe drives, each directly connected, for 2TB of raw capacity with no shared storage. If the drives are assigned separately to application data and backups, define their roles clearly. The second drive in the same server is not a substitute for an off-site backup.

Suitable for administrators with external backups who want to plan separate volumes. Not suitable when two independent drives are expected to provide redundancy.

Usable space varies with formatting and filesystem overhead

Prioritize redundancy: configure a mirror

An administrator can configure a ZFS mirror or software RAID across the two 1TB drives. A two-drive mirror provides about 1TB of nominal usable capacity, so data cannot be planned around the raw capacity of both drives.

Suitable for teams able to manage a storage pool, monitor alerts and verify recovery. Not suitable when data center-managed arrays, backups and recovery policies are expected.

No hardware RAID; mirroring is not a backup
Translate the requirements into five choices on the configuration page
OptionAvailable choicesWhat to consider before ordering
Memory128GB or 256GB DDR4-3200 ECCChoose based on service working sets, VM allocations and ZFS ARC requirements. Both options use an AMD EPYC 7543P with 32 cores and 64 threads.
Additional drivesAdd 0–2 × 2TB NVMe drivesRaw capacity totals 2TB with no added drives, 4TB with one, or 6TB with two. Plan separately for storage pools and mirrored capacity.
Bandwidth250 Mbps, 1 Gbps or 2 GbpsAll three options include unlimited traffic. Choose a rate based on data synchronization volume and recovery windows; port speed does not guarantee download speeds from arbitrary sources.
IPv4 addresses1 included; up to 255 additional addressesConfirm service bindings and VM address requirements first. Use the first address for management access and configure any additional addresses in the operating system.
Operating system14 free operating systems, plus paid Windows ServerCheck software compatibility and access requirements first. Windows Server is licensed monthly, with English, Traditional Chinese and Simplified Chinese available as system languages.
02 · Payment

Check the monthly total, then confirm the amount due

AMD dedicated servers are available on monthly, quarterly, semi-annual and annual billing cycles, covering 1, 3, 6 or 12 months. The cycle determines the number of months covered; it does not change the configuration or include a discount. Billing is in US dollars (USD), with no setup fee and no contract.

How billing is calculated Amount due = monthly total × number of months

The monthly total includes the server's monthly price, additional drives, bandwidth upgrades, extra IPv4 addresses and any selected Windows license. Unselected add-ons are not included.

  1. 01

    Review the configuration

    Check memory, drive count, bandwidth, number of addresses and operating system to avoid overlooking add-on charges when comparing base prices.

  2. 02

    Confirm the billing period

    Make sure the amount due matches the selected cycle. The monthly total is not the full amount due for a quarterly, semi-annual or annual cycle.

  3. 03

    Choose a listed payment method

    Payment methods are those listed on the configuration page before payment is confirmed. After payment, refer to the status returned with the order.

03 · Delivery

Check the control panel once the server is ready

Payment and server preparation are separate steps. After payment, check the order in the OnuCloud control panel. Once the server is ready, its status and login credentials appear in the server details. Closing the payment page does not remove the delivery record and is not a reason to place a duplicate order.

Match the server to its order

Compare the server details with the order, especially the memory option, drives, bandwidth, IPv4 addresses and operating system. For multiple orders, record the purpose of each server separately to avoid mixing production credentials with test systems.

Check delivery details in OnuCloud

Use credentials only for authorized administration

After delivery, retrieve the address and login credentials from the control panel. Usernames and ports are not predetermined. Do not put credentials in public code, screenshots or shared documents. When submitting diagnostic details, keep passwords out of email subject lines and public content.

Read the OnuCloud first-connection guide

For questions about status or configuration, submit a support ticket in the control panel with the order number, displayed status and a description of the issue. Availability is based on the status reported by the control panel in real time.

04 · First login

Verify storage, networking and addresses before starting workloads

Linux, FreeBSD and Proxmox VE use SSH; Windows Server uses RDP. On servers with multiple IPv4 addresses, connect using the first address. Refer to the credentials in the control panel delivery record. Detailed connection steps are on the connection guide page; this page covers only the provisioning checklist.

  1. 01

    Identify each drive

    Confirm that the system recognizes both 1TB NVMe drives separately and that the number of added drives matches the order. Verify device identities before creating a pool; do not format drives based only on their device order.

  2. 02

    Test network connectivity

    Check management access and the network paths required by the workloads. If additional IPv4 addresses are needed, bind them in the operating system and verify service listeners and firewall rules.

  3. 03

    Set up storage and backups

    Create filesystems, a ZFS mirror or software RAID according to the plan, then configure off-site backups. Before migrating production data, test recovery with a small sample.

  4. 04

    Secure administrative access

    Change the administrative password after first login, configure SSH keys as needed and restrict unnecessary access. Record drive assignments, address allocation and recovery steps before scheduling synchronization jobs.

Responsibilities

Assign separate owners for hardware, networking and system data

Direct-attached drives provide independent storage devices; they are not a managed storage service. Before provisioning, decide who will investigate failures and where data will be restored from. That matters more than recording capacity alone.

OnuCloud: hardware and data center networking

  • Server hardwareHardware-related faults are investigated and coordinated through support tickets.
  • Drive replacementHardware handling is provided for failed drives. Replacing a drive does not restore data from the original drive.
  • Data center networkData center network issues on the service side are investigated using the order details and reported symptoms.
Include the order number, symptoms and checks already completed when reporting an issue

Customer: system configuration and data protection

  • System and applicationsManage accounts, access rules, software updates, service configuration and address bindings.
  • Data redundancyDeploy and maintain software RAID or a ZFS mirror, and check storage pool health.
  • Off-site backupsSet backup and recovery policies, keep a separate copy and verify that recovery works.
No hardware RAID; customers are responsible for data backups
Changes after provisioning

Changing memory requires a new order; back up data before reinstalling

Memory cannot be upgraded on the original server after provisioning. Switching between 128GB and 256GB requires ordering another server and planning data migration and workload cutover. A new order does not add memory directly to the existing server.

  1. 01

    Change the memory option

    Choose the new server configuration first, keep a recoverable copy of the data and then plan the migration. Check the orders and credentials for both servers separately during the migration.

  2. 02

    Reinstall the operating system

    Use a control panel support ticket to specify the order, target operating system and requirements. Back up application data and configuration beforehand; do not rely on copies on the local drives for recovery after reinstallation.

  3. 03

    Report a hardware fault

    Describe unrecognized devices, system errors or connection symptoms. Hardware handling and data recovery are separate processes; customers remain responsible for backups and recovery.

Renewal and cancellation

Renew for the selected billing cycle and plan data migration before expiry

AMD dedicated servers renew according to the selected billing cycle, with no contract. If not renewed, service stops at the end of the cycle. Decide whether to continue based on the workload plan and data migration progress, not just whether a backup job has run once.

Continue service

Check the billing cycle and expiry details in the control panel, plan renewal in advance and keep verifying backup results. Renewal extends the service period; it does not replace system administration or backup checks.

Refer to the order record

End service

Before expiry, move data, application configuration and recovery materials to a separate location. Verify that copies are readable and recovery steps work before removing dependencies on the server.

Do not rely on access to data on the original server after service ends

DGX Spark has a separate ordering process

See the DGX rental guide for OnuCloud DGX Spark provisioning and end-of-service data rules. The AMD configuration process on this page does not apply.

Turn storage plans into an order

Once capacity, redundancy and backup responsibilities are clear, submit a configuration. For existing orders, check the status in the control panel.