IBM Cloud Docs
Bare metal server options

Bare metal server options

Your IBM Cloud® bare metal server is an hourly or monthly, single-tenant server that is dedicated to you. Your server isn't shared in any part, including server resources, with other customers. You manage your server, which is provisioned without a hypervisor, and deployed in one or more data centers. Multiple Bare Metal Servers can communicate on the IBM Cloud virtual private network as if stationed on the same rack.

Servers for every workload

IBM Cloud has Bare Metal Servers to fit every workload. For more information, see Bare metal servers.

Custom-based servers

If a fast-provisioning server doesn't meet your workload needs, you can customize your Bare Metal Servers to meet your needs. Customized servers are generally provisioned in 2 - 4 hours. The provisioning time depends on complexity, quantity, and testing options.

SAP-certified bare metal servers

IBM Cloud Bare Metal Servers are certified to support your SAP HANA and SAP NetWeaver workloads. For more information, see SAP-certified infrastructure.

VMware-certified servers

IBM Cloud® provides the capability for you to provision dedicated Bare Metal Servers so you can deploy your own VMware®-based private cloud. For more information, see Getting started with VMware.

Advanced hardware options

IBM Cloud offers advanced hardware options to fit your workload needs.

Intel Sapphire Rapids CPUs

Intel® Sapphire Rapids processors are available only in the DAL, MAD, TOK, TOR, and WDC regions.

You can now choose the following Sapphire Rapids CPUs when you provision a bare metal server.

Sapphire Rapids CPU options
Sapphire Rapids CPU Specifications Memory TPM 2.0 support
8474C 48 core, 2.1 GHz 256 GB, 512 GB, 1024 GB, 2048 GB, 4096 GB Yes

Sapphire Rapids processors support the following operating systems:

  • Debian 11
  • Microsoft Windows 19 & 2022
  • RHEL 9.x
  • Rocky Linux
  • Ubuntu 22.04
  • No OS option

AMD CPUs

AMD EPYC® "Rome" and "Milan" generation CPUs are now an option. Rome generation processors are high-performance multiprocessors that are based on AMD's Zen 2 architecture. The EPYC processors offer several CPU options that offer up to 64 cores per socket.

EPYC processors are available in select data centers.

You can choose from the following AMD EPYC "Rome" and "Milan" CPUs when you provision a bare metal server:

AMD EPYC CPU options
AMD EPYC CPUs Specifications
AMD EPYC 7642 48-core, 2.3 GHz
AMD EPYC 7F72 24-core, 3.2 GHz
AMD EPYC 7763 64-core, 2.45 GHz

The following operating systems are supported by AMD EPYC "Rome" CPUs:

  • CentOS 7.9
  • Microsoft® Server 2019
  • RHEL 7.x, 8.x
  • Ubuntu 18.04

Systems must be started in UEFI mode.

For more information about AMD EPYC CPUs on IBM Cloud®, see the AMD on IBM Cloud® Bare Metal Servers.

Intel Cascade Lake CPUs

You can choose from the following Intel Xeon® Cascade Lake CPUs when you provision a bare metal server:

Cascade Lake CPU options
Cascade Lake CPUs Specifications
Intel Xeon 4210 10-Core, 2.2 GHz
Intel Xeon 5218 16-Core, 2.3 GHz
Intel Xeon 6248 20-Core, 2.6 GHz
Intel Xeon 6250 8-Core, 3.9 GHz
Intel Xeon 8260 48-core, 2.4 Ghz

NVIDIA GPUs

For certain bare metal servers, you can add the processing power of NVIDIA® GPUs. When you select a bare metal server, look for GPU in the Features column on the provisioning page. Make sure that you install the appropriate drivers. See NVIDIA drivers

Server enhancement options

When you provision a bare metal server, you have the following enhancement options to help make managing your server easier. Keep in mind that these options might vary depending on your server configuration.

Dynamic inventory

You can now see what servers are available in what data center when you provision a bare metal server. If a server is not available in the data center you selected, hover over the server name. A list is displayed that indicates the data centers in which the server is available. For more information about provisioning a bare metal server, see Selecting from fast provisioning servers.

Network redundancy

Port redundancy provides a networking failover by maintaining a primary and secondary network port. If the primary port fails, the secondary (redundant) port enables.

Only one port is active at a time.

The following network redundancy options are available for bare metal servers.

Network redunancy options
Redundancy options Description
Automatic redundancy (Recommended option) Automatically configures the redundant ports for interface teaming through LACP (Link Aggregation Control Protocol) to preserve connectivity during routine maintenance.
User-managed redundancy Interface teaming must be configured on the host operating system to use network redundancy. Without interface teaming, connectivity during routine maintenance is not preserved.
No redundancy This option is not recommended.
Interface teaming (link aggregation) Combines, in parallel, multiple network connections to provide redundancy and increase network throughput.

VMWare requires user-managed links.

For more information about network options, see Network options.

Block and file storage add-on

If you need extra storage, IBM makes it easy! You can now order block and file storage (20 - 12,000 GB) when you provision a bare metal server.

Your add-on storage isn't automatically connected to your bare metal server. You need to connect the add-on storage to your bare metal server after your server provisions.

For more information about block and file storage, see the following links.

Bare metal server add-ons

The following add-ons are available when you provision a bare metal server.

Bare metal server provisioning options
Option Description
Power supply You can provision your bare metal server with two independent power supply units. This redundancy within the data center helps maintain uptime during unplanned or planned electrical maintenance.
IBM Cloud® Backup IBM Cloud® Backup is an automated, agent-based backup and recovery system that is managed through the Cloud Backup WebCC browser utility. For more information, see Getting started with IBM Cloud® Backup.
Server security Intel Trusted Execution Technology (TXT) provides hardware-assisted security technologies to enhance your security portfolio and act as an extra security for your infrastructure.
Intel Software Guard Extensions (SGX) Intel SGX is used to partition sensitive data into secure enclaves by using a set of security-related application code. You can add SGX when you provision your bare metal server.
Business continuance insurance (BCI) Business continuance insurance helps you avoid overage charges if you experience a network attack (DDOS) that uses all of your allowed bandwidth.
Firewall A hardware firewall provides an extra layer of security that is provisioned on demand without service interruptions. This firewall prevents unwanted traffic from reaching your servers by reducing your attack surface, and by enabling your server resources to be dedicated for their intended use. For more information, see Getting started with Hardware Firewall.
Microsoft SQL Server A relational database management system that manages data storage and retrieval. For more information, see SQL Server 2019 and SQL Server 2022.
Monitoring - host ping Basic monitoring is used to initiate service and slow pings to make sure that the device is online and responsive. For more information, see Basic monitoring.
Notification You can define an alert on a single metric or a set of metrics to notify you of events or issues that you want to monitor. For more information, see Working with alerts.
Response Automated response to your monitored metrics notifications.
Public secondary IP addresses You can request extra IP addresses for your server, which is recommended if you announce services externally. For more information about secondary IP addresses, see Secondary subnets. If you need extra IP addresses later, you can always order extra secondary subnets. For more information, see Ordering secondary subnets and global IP addresses.
IPv6 IP addresses An IPv6 is the most recent numeric label that is used to identify and locate a network interface of a computer or a network node that participates in a computer network by using the IPv6 protocol. A single IPv6 address is included.