Ask Your Question
0

management network interface VS provider network interface

asked 2019-06-07 18:53:01 -0500

najeh gravatar image

An openstack node should have two NIC's a mangement one and a provider one.

what is the difference between these two interfaces and the role of each one ?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
1

answered 2019-06-07 20:06:23 -0500

updated 2019-06-07 20:13:06 -0500

I would not say that this is correct. Think of all the network traffic in a cloud: Between VMs (via tunnels or VLANs), between VMs and the outside world, iSCSI, external APIs, internal APIs, message queue, database, Swift data, Ceph data, Ceph control, live migration, and whatever else may come to mind.

Depending on the size of your installation, you may want to separate this traffic into several networks. However, nodes don't usually have five or ten network interfaces (except for blade servers, where network interfaces can be added at will). Typical example: Networks are implemented as VLANs, and nodes have a single interface bond that carries all VLANs.

Here is a blog series with a more realistic description of production networking. It was written at the Liberty/Mitaka timeframe, but the principles are the same today.

Of course, for a proof-of-concept cloud, training, self-education etc., you can set up OpenStack nodes with two networks and interfaces. But you don't need two. A single network is sufficient, in particular for an all-in-one cloud.

To answer your question

The provider interface connects the node to a provider network, such as an external network. This gives instances connectivity to the outside world.

The management network carries API traffic between the OpenStack servers, as well as message queue, database etc. traffic. In a two-network configuration, it probably also carries VM traffic over GRE or VXLAN tunnels, and external API traffic. In short, everything except provider traffic.

edit flag offensive delete link more

Comments

@Bernd Bausch the management interface present the public IP or the private IP?

why we use up and down in the configuration of the provider interface ?

najeh gravatar imagenajeh ( 2019-06-16 10:09:16 -0500 )edit
1

The management interface has nothing to do with instance IP addresses.

Public IP addresses come from the network to which the provider interface is connected. Private IP addresses are from the virtual networks that are owned by projects.

Bernd Bausch gravatar imageBernd Bausch ( 2019-06-16 14:29:47 -0500 )edit

@Bernd Bausch why we use up and down in the configuration of the provider interface ?

najeh gravatar imagenajeh ( 2019-06-16 15:57:44 -0500 )edit
1

What do you mean by "up and down"? And which configuration?

Perhaps this: The provider interface doesn't need an IP address if it is only used for provider traffic. But I think it should be up.

Bernd Bausch gravatar imageBernd Bausch ( 2019-06-17 02:25:16 -0500 )edit

@Bernd Bausch This configuration :

The provider network interface

auto INTERFACE_NAME

iface INTERFACE_NAME inet manual

up ip link set dev $IFACE up

down ip link set dev $IFACE down

najeh gravatar imagenajeh ( 2019-06-17 02:49:55 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2019-06-07 18:53:01 -0500

Seen: 53 times

Last updated: Jun 07