Chapter 2. Hardware requirements

This section describes the hardware requirements for NFV.

For a complete list of the certified hardware for Red Hat OpenStack Platform, see Red Hat OpenStack Platform certified hardware.

2.1. Tested NICs

For a list of tested NICs for NFV, see the Red Hat Knowledgebase solution Network Adapter Fast Datapath Feature Support Matrix.

If you configure OVS-DPDK on Mellanox ConnectX-4 or ConnectX-5 network interfaces, you must set the corresponding kernel driver in the compute-ovs-dpdk.yaml file:

members
 - type: ovs_dpdk_port
    name: dpdk0
    driver: mlx5_core
    members:
    - type: interface
      name: enp3s0f0

2.2. Discovering your NUMA node topology

When you plan your deployment, you must understand the NUMA topology of your Compute node to partition the CPU and memory resources for optimum performance. To determine the NUMA information, perform one of the following tasks:

  • Enable hardware introspection to retrieve this information from bare-metal nodes.
  • Log on to each bare-metal node to manually collect the information.
Note

You must install and configure the undercloud before you can retrieve NUMA information through hardware introspection. For more information about undercloud configuration, see: Director Installation and Usage Guide.

Retrieving hardware introspection details

The Bare Metal service hardware-inspection-extras feature is enabled by default, and you can use it to retrieve hardware details for overcloud configuration. For more information about the inspection_extras parameter in the undercloud.conf file, see Configuring the Director.

For example, the numa_topology collector is part of the hardware-inspection extras and includes the following information for each NUMA node:

  • RAM (in kilobytes)
  • Physical CPU cores and their sibling threads
  • NICs associated with the NUMA node

To retrieve the information listed above, substitute <UUID> with the UUID of the bare-metal node to complete the following command:

# openstack baremetal introspection data save <UUID> | jq .numa_topology

The following example shows the retrieved NUMA information for a bare-metal node:

{
  "cpus": [
    {
      "cpu": 1,
      "thread_siblings": [
        1,
        17
      ],
      "numa_node": 0
    },
    {
      "cpu": 2,
      "thread_siblings": [
        10,
        26
      ],
      "numa_node": 1
    },
    {
      "cpu": 0,
      "thread_siblings": [
        0,
        16
      ],
      "numa_node": 0
    },
    {
      "cpu": 5,
      "thread_siblings": [
        13,
        29
      ],
      "numa_node": 1
    },
    {
      "cpu": 7,
      "thread_siblings": [
        15,
        31
      ],
      "numa_node": 1
    },
    {
      "cpu": 7,
      "thread_siblings": [
        7,
        23
      ],
      "numa_node": 0
    },
    {
      "cpu": 1,
      "thread_siblings": [
        9,
        25
      ],
      "numa_node": 1
    },
    {
      "cpu": 6,
      "thread_siblings": [
        6,
        22
      ],
      "numa_node": 0
    },
    {
      "cpu": 3,
      "thread_siblings": [
        11,
        27
      ],
      "numa_node": 1
    },
    {
      "cpu": 5,
      "thread_siblings": [
        5,
        21
      ],
      "numa_node": 0
    },
    {
      "cpu": 4,
      "thread_siblings": [
        12,
        28
      ],
      "numa_node": 1
    },
    {
      "cpu": 4,
      "thread_siblings": [
        4,
        20
      ],
      "numa_node": 0
    },
    {
      "cpu": 0,
      "thread_siblings": [
        8,
        24
      ],
      "numa_node": 1
    },
    {
      "cpu": 6,
      "thread_siblings": [
        14,
        30
      ],
      "numa_node": 1
    },
    {
      "cpu": 3,
      "thread_siblings": [
        3,
        19
      ],
      "numa_node": 0
    },
    {
      "cpu": 2,
      "thread_siblings": [
        2,
        18
      ],
      "numa_node": 0
    }
  ],
  "ram": [
    {
      "size_kb": 66980172,
      "numa_node": 0
    },
    {
      "size_kb": 67108864,
      "numa_node": 1
    }
  ],
  "nics": [
    {
      "name": "ens3f1",
      "numa_node": 1
    },
    {
      "name": "ens3f0",
      "numa_node": 1
    },
    {
      "name": "ens2f0",
      "numa_node": 0
    },
    {
      "name": "ens2f1",
      "numa_node": 0
    },
    {
      "name": "ens1f1",
      "numa_node": 0
    },
    {
      "name": "ens1f0",
      "numa_node": 0
    },
    {
      "name": "eno4",
      "numa_node": 0
    },
    {
      "name": "eno1",
      "numa_node": 0
    },
    {
      "name": "eno3",
      "numa_node": 0
    },
    {
      "name": "eno2",
      "numa_node": 0
    }
  ]
}

2.3. BIOS Settings

The following table describes the required BIOS settings for NFV:

Note

You must enable SR-IOV global and NIC settings in the BIOS, or your Red Hat OpenStack Platform (RHOSP) deployment with SR-IOV Compute nodes will fail.

Table 2.1. BIOS Settings

ParameterSetting

C3 Power State

Disabled.

C6 Power State

Disabled.

MLC Streamer

Enabled.

MLC Spacial Prefetcher

Enabled.

DCU Data Prefetcher

Enabled.

DCA

Enabled.

CPU Power and Performance

Performance.

Memory RAS and Performance Config → NUMA Optimized

Enabled.

Turbo Boost

Disabled in NFV deployments that require deterministic performance.
Enabled in all other scenarios.

VT-d

Enabled for Intel cards if VFIO functionality is needed.

NUMA memory interleave

Disabled

On processors that use the intel_idle driver, Red Hat Enterprise Linux can ignore BIOS settings and re-enable the processor C-state.

You can disable intel_idle and instead use the acpi_idle driver by specifying the key-value pair intel_idle.max_cstate=0 on the kernel boot command line.

Confirm that the processor is using the acpi_idle driver by checking the contents of current_driver:

# cat /sys/devices/system/cpu/cpuidle/current_driver
acpi_idle
Note

You will experience some latency after changing drivers, because it takes time for the Tuned daemon to start. However, after Tuned loads, the processor does not use the deeper C-state.