Files
openstack-ansible/doc/source/reference/architecture/metal-networking.rst
Ivan Anfimov fda1d01b41 docs: replace "tenant" to "project"
Change-Id: I6c0d36317ff76f45dfa8f4b380ffa6de2ad54e6a
Signed-off-by: Ivan Anfimov <lazekteam@gmail.com>
2025-11-23 11:01:10 +00:00

144 lines
5.1 KiB
ReStructuredText

.. _metal-networking:
Metal networking
====================
OpenStack-Ansible supports deploying OpenStack and related services on "metal"
as well as inside LXC containers. Python virtual environments (venvs) provide
OpenStack service and Python library segregation, while other services such
as Galera and RabbitMQ are co-located on the host. All services in this
deployment model share the *same* IP address.
This appendix describes how the interfaces are connected and how traffic flows.
For more information about how the OpenStack Networking service (Neutron) uses
the interfaces for instance traffic, please see the
`OpenStack Networking Guide`_.
.. _OpenStack Networking Guide: https://docs.openstack.org/neutron/latest/admin/index.html
For details on the configuration of networking for your
environment, please have a look at :ref:`openstack-user-config-reference`.
Physical host interfaces
~~~~~~~~~~~~~~~~~~~~~~~~
In a typical production environment, physical network interfaces are combined
in bonded pairs for better redundancy and throughput. Avoid using two ports on
the same multiport network card for the same bonded interface, because a
network card failure affects both of the physical network interfaces used by
the bond. Multiple bonded interfaces (ie. bond0, bond1) can be used to
segregate traffic, if desired. Single (bonded) interfaces are also a supported
configuration, but will require the use of VLAN subinterfaces.
Linux Bridges/Switches
~~~~~~~~~~~~~~~~~~~~~~
The combination of containers and flexible deployment options requires
implementation of advanced Linux networking features, such as bridges,
switches, and namespaces.
* Bridges provide layer 2 connectivity (similar to switches) among
physical, logical, and virtual network interfaces within a host. After
a bridge/switch is created, the network interfaces are virtually plugged
in to it.
OpenStack-Ansible uses Linux bridges for control plane connections to LXC
containers, and can use Linux bridges or Open vSwitch-based bridges for
data plane connections that connect virtual machine instances to the
physical network infrastructure.
* Network namespaces provide logically separate layer 3 environments (similar
to VRFs) within a host. Namespaces use virtual interfaces to connect
with other namespaces, including the host namespace. These interfaces,
often called ``veth`` pairs, are virtually plugged in between
namespaces similar to patch cables connecting physical devices such as
switches and routers.
Network diagrams
~~~~~~~~~~~~~~~~
Hosts with services running on metal
------------------------------------
The following diagram shows how all of the interfaces and bridges interconnect
to provide network connectivity to the OpenStack deployment:
.. image:: ../figures/networkarch-metal-external.drawio.png
Neutron traffic
---------------
Common reference drivers, including ML2/OVS, and ML2/OVN, and their
respective agents, are responsible for managing the virtual networking
infrastructure on each node. OpenStack-Ansible refers to Neutron traffic
as "data plane" traffic, and can consist of flat, VLAN, or overlay technologies
such as VXLAN and Geneve.
Neutron agents can be deployed across a variety of hosts, but are typically
limited to dedicated network hosts or infrastructure hosts (controller nodes).
Neutron agents are deployed "on metal" and not within an LXC container. Neutron
typically requires the operator to define "provider bridge mappings", which map
a provider network name to a physical interface. These provider bridge mappings
provide flexibility and abstract physical interface names when creating provider
networks.
**Open vSwitch/OVN Example**:
.. code-block:: ini
bridge_mappings = physnet1:br-ex
OpenStack-Ansible provides two overrides when defining provider networks that
can be used for creating the mappings and in some cases, connecting the physical
interfaces to provider bridges:
- ``host_bind_override``
- ``network_interface``
The ``host_bind_override`` key is used to replace an LXC-related interface
name with a physical interface name when a component is deployed on bare metal hosts.
It will be used to populate ``network_mappings`` for Neutron.
The ``network_interface`` override is used for Open vSwitch and OVN-based deployments,
and requires a physical interface name which will be connected to the provider bridge
(ie. br-ex) for flat and vlan-based provider and project network traffic.
The following diagrams reflect the differences in the virtual network layout for
supported network architectures.
Open vSwitch (OVS)
..................
Networking Node
***************
.. image:: ../figures/networking-openvswitch-nn.drawio.png
:align: center
Compute Node
************
.. image:: ../figures/networking-openvswitch-cn.drawio.png
:align: center
Open Virtual Network (OVN)
..........................
.. note::
The ML2/OVN mechanism driver is deployed by default
as of the Zed release of OpenStack-Ansible.
Networking Node
***************
.. image:: ../figures/networking-ovn-nn.drawio.png
:align: center
Compute Node
************
.. image:: ../figures/networking-ovn-cn.drawio.png
:align: center