proxmox-mirrors/pve-docs
1
0
Fork 0
mirror of https://git.proxmox.com/git/pve-docs synced 2025-06-26 06:29:29 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

rework SDN docs a bit

Browse Source 
drop thumbnails for now, would need updating anyway and the examples
are of bigger help for now anyway, IMO.

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This commit is contained in:
Thomas Lamprecht 2020-05-10 19:43:36 +02:00
parent 1556b768a6
commit ee6e18c480
1 changed files with 261 additions and 199 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

414
pvesdn.adoc
View File

@ -5,238 +5,256 @@ ifndef::manvolnum[]
:pve-toplevel:
endif::manvolnum[]
The SDN feature allow to create virtual networks (vnets)
at datacenter level.
The **S**oftware **D**efined **N**etwork (SDN) feature allows one to create
virtual networks (vnets) at datacenter level.
To enable SDN feature, you need to install "libpve-network-perl" package
WARNING: SDN is currently an **experimental feature** in {pve}. This
Documentation for it is also still under development, ask on our
xref:getting_help[mailing lists or in the forum] for questions and feedback.
Installation
------------
To enable the experimental SDN integration, you need to install
"libpve-network-perl" package
----
apt install libpve-network-perl
----
A vnet is a bridge with a vlan or vxlan tag.
The vnets are deployed locally on each node after configuration
commit at datacenter level.
You need to have "ifupdown2" package installed on each node to manage local
configuration reloading.
You need to have `ifupdown2` package installed on each node to manage local
configuration reloading without reboot:
----
apt install ifupdown2
----
Basic Overview
--------------
The {pve} SDN allows separation and fine grained control of Virtual Guests
networks, using flexible software controlled configurations.
Separation consists of zones, a zone is it's own virtual separated area.
A Zone can be used by one or more 'VNets'. A 'VNet' is virtual network in a
zone. Normally it shows up as a common Linux bridge with either a VLAN or
'VXLAN' tag, or using layer 3 routing for control.
The 'VNets' are deployed locally on each node, after configuration was commited
from the cluster wide datacenter level.
Main configuration
------------------
The configuration is done at datacenter level.
The configuration is done at datacenter (cluster-wide) level, it will be saved
in configuration files located in the shared configuration file system:
`/etc/pve/sdn`
The sdn feature have 4 main sections for the configuration
On the web-interface SDN feature have 4 main sections for the configuration
* SDN
* SDN: a overview of the SDN state
* Zones
* Zones: Create and manage the virtual separated network Zones
* Vnets
* VNets: The per-node building block to provide a Zone for VMs
* Controller
* Controller:
SDN
~~~
[thumbnail="screenshot/gui-sdn-status.png"]
This is the main status panel. Here you can see deployment status of zones on
different nodes.
This is the Main panel, where you can see deployment of zones on differents nodes.
They are an "apply" button, to push && reload local configuration on differents nodes.
There is an 'Apply' button, to push and reload local configuration on all
cluster nodes nodes.
Zones
~~~~~
[thumbnail="screenshot/gui-sdn-zone.png"]
A zone will define a virtually separated network.
A zone will defined the kind of virtual network you want to defined.
It can use different technologies for separation:
it can be
* VLAN: Virtual LANs are the classic method to sub-divide a LAN
* vlan
* QinQ: stacked VLAN (formally known as `IEEE 802.1ad`)
* QinQ (stacked vlan)
* VXLAN: (layer2 vxlan)
* vxlan (layer2 vxlan)
* bgp-evpn (vxlan with layer3 routing)
* bgp-evpn: vxlan using layer3 border gateway protocol routing
You can restrict a zone to specific nodes.
It's also possible to add permissions on a zone, to restrict user
to use only a specific zone and the vnets in this zone
It's also possible to add permissions on a zone, to restrict user to use only a
specific zone and only the VNets in that zone
Vnets
VNets
~~~~~
[thumbnail="screenshot/gui-sdn-vnet-evpn.png"]
A `VNet` is in its basic form just a Linux bridge that will be deployed locally
on the node and used for Virtual Machine communication.
A vnet is a bridge that will be deployed locally on the node,
for vm communication. (Like a classic vmbrX).
VNet properties are:
Vnet properties are:
* ID: a 8 characters ID to name and identify a VNet
* ID: a 8 characters ID
* Alias: Optional longer name, if the ID isn't enough
* Alias: Optionnal bigger name
* Zone: The associated zone for this VNet
* Zone: The associated zone of the vnet
* Tag: The unique VLAN or VXLAN id
* Tag: unique vlan or vxlan id
* IPv4: an anycast IPv4 address, it will be configured on the underlying bridge
on each node part of the Zone. It's only useful for `bgp-evpn` routing.
* ipv4: an anycast ipv4 address (same bridge ip deployed on each node), for bgp-evpn routing only
* ipv6: an anycast ipv6 address (same bridge ip deployed on each node), for bgp-evpn routing only
* IPv6: an anycast IPv6 address, it will be configured on the underlying bridge
on each node part of the Zone. It's only useful for `bgp-evpn` routing.
Controllers
~~~~~~~~~~~
[thumbnail="screenshot/gui-sdn-controller.png"]
Some zone plugins (Currently : bgp-evpn only),
need an external controller to manage the vnets control-plane.
Some zone types (currently only the `bgp-evpn` plugin) need an external
controller to manage the VNet control-plane.
Zones Plugins
-------------
common zone options:
* nodes: restrict deploy of the vnets of theses nodes only
Common options
~~~~~~~~~~~~~~
nodes:: deploy and allow to use a VNets configured for this Zone only on
these nodes.
Vlan
~~~~~
VLAN Zones
~~~~~~~~~~
[thumbnail="screenshot/gui-sdn-zone-vlan.png"]
This is the simplest plugin, it will reuse an existing local Linux or OVS
bridge, and manage VLANs on it.
The benefit of using SDN module, is that you can create different zones with
specific VNets VLAN tag, and restrict Virtual Machines to separated zones.
This is the most simple plugin, it'll reuse an existing local bridge or ovs,
and manage vlan on it.
The benefit of using sdn module, is that you can create different zones with specific
vnets vlan tag, and restrict your customers on their zones.
Specific `VLAN` configuration options:
specific qinq configuration options:
bridge:: Reuse this local VLAN-aware bridge, or OVS interface, already
configured on *each* local node.
* bridge: a local vlan-aware bridge or ovs switch already configured on each local node
QinQ Zones
~~~~~~~~~~
QinQ
~~~~~
QinQ is stacked VLAN. The first VLAN tag defined for the zone
(so called 'service-vlan'), and the second VLAN tag defined for the vnets
[thumbnail="screenshot/gui-sdn-zone-qinq.png"]
NOTE: Your physical network switchs must support stacked VLANs!
QinQ is stacked vlan.
you have the first vlan tag defined on the zone (service-vlan), and
the second vlan tag defined on the vnets
Specific QinQ configuration options:
Your physical network switchs need to support stacked vlans !
bridge:: a local VLAN-aware bridge already configured on each local node
service vlan:: he main VLAN tag of this zone
mtu:: Due to the double stacking of tags you need 4 more bytes for QinQ VLANs.
For example, you reduce the MTU to `1496` if you physical interface MTU is
`1500`.
specific qinq configuration options:
VXLAN Zones
~~~~~~~~~~~
* bridge: a local vlan-aware bridge already configured on each local node
* service vlan: The main vlan tag of this zone
* mtu: you need 4 more bytes for the double tag vlan.
You can reduce the mtu to 1496 if you physical interface mtu is 1500.
The VXLAN plugin will establish a tunnel (named overlay) on top of an existing
network (named underlay). It encapsulate layer 2 Ethernet frames within layer
4 UDP datagrams, using `4789` as the default destination port. You can, for
example, create a private IPv4 VXLAN network on top of public internet network
nodes.
This is a layer2 tunnel only, no routing between different VNets is possible.
Vxlan
~~~~~
Each VNet will have use specific VXLAN id from the range (1 - 16777215).
[thumbnail="screenshot/gui-sdn-zone-vxlan.png"]
Specific EVPN configuration options:
The vxlan plugin will established vxlan tunnel (overlay) on top of an existing network (underlay).
you can for example, create a private ipv4 vxlan network on top of public internet network nodes.
This is a layer2 tunnel only, no routing between different vnets is possible.
peers address list:: a list of IPs from all nodes where you want to communicate (can also be external nodes)
mtu:: because VXLAN encapsulation use 50bytes, the MTU need to be 50 bytes lower than the outgoing physical interface.
Each vnet will have a specific vxlan id ( 1 - 16777215 )
EVPN Zones
~~~~~~~~~~
This is the most complex of all supported plugins.
Specific evpn configuration options:
BGP-EVPN allows one to create routable layer3 network. The VNet of EVPN can
have an anycast IP-address and or MAC-address. The bridge IP is the same on each
node, with this a virtual guest can use that address as gateway.
* peers address list: an ip list of all nodes where you want to communicate (could be also external nodes)
Routing can work across VNets from different zones through a VRF (Virtual
Routing and Forwarding) interface.
* mtu: because vxlan encapsulation use 50bytes, the mtu need to be 50 bytes lower
than the outgoing physical interface.
Specific EVPN configuration options:
evpn
~~~~
VRF VXLAN Tag:: This is a vxlan-id used for routing interconnect between vnets,
it must be different than VXLAN-id of VNets
[thumbnail="screenshot/gui-sdn-zone-evpn.png"]
controller:: an EVPN-controller need to be defined first (see controller
plugins section)
This is the most complex plugin.
BGP-evpn allow to create routable layer3 network.
The vnet of evpn can have an anycast ip address/mac address.
The bridge ip is the same on each node, then vm can use
as gateway.
The routing is working only across vnets of a specific zone through a vrf.
Specific evpn configuration options:
* vrf vxlan tag: This is a vxlan-id used for routing interconnect between vnets,
it must be different than vxlan-id of vnets
* controller: an evpn need to be defined first (see controller plugins section)
* mtu: because vxlan encapsulation use 50bytes, the mtu need to be 50 bytes lower
than the outgoing physical interface.
mtu:: because VXLAN encapsulation use 50bytes, the MTU need to be 50 bytes
lower than the outgoing physical interface.
Controllers Plugins
-------------------
evpn
~~~~
EVPN Controller
~~~~~~~~~~~~~~~
[thumbnail="screenshot/gui-sdn-controller-evpn.png"]
For bgp-evpn, we need a controller to manage the control plane.
The software controller is "frr" router.
You need to install it on each node where you want to deploy the evpn zone.
For `BGP-EVPN`, we need a controller to manage the control plane.
The currently supported software controller is the "frr" router.
You may need to install it on each node where you want to deploy EVPN zones.
----
apt install frr
----
configuration options:
Configuration options:
*asn: a unique bgp asn number.
It's recommended to use private asn number (64512 – 65534, 4200000000 – 4294967294)
asn:: a unique BGP ASN number. It's highly recommended to use private ASN
number (64512 – 65534, 4200000000 – 4294967294), as else you could end up
breaking, or get broken, by global routing by mistake.
*peers: an ip list of all nodes where you want to communicate (could be also external nodes or route reflectors servers)
peers:: an ip list of all nodes where you want to communicate (could be also
external nodes or route reflectors servers)
If you want to route traffic from the sdn bgp-evpn network to external world:
Additionally, if you want to route traffic from a SDN BGP-EVPN network to
external world:
* gateway-nodes: The proxmox nodes from where the bgp-evpn traffic will exit to external through the nodes default gateway
gateway-nodes:: The proxmox nodes from where the bgp-evpn traffic will exit to
external through the nodes default gateway
If you want that gateway nodes don't use the default gateway, but for example, sent traffic to external bgp routers
If you want that gateway nodes don't use the default gateway, but, for example,
sent traffic to external BGP routers
* gateway-external-peers: 192.168.0.253,192.168.0.254
gateway-external-peers:: 192.168.0.253,192.168.0.254
Local deployment Monitoring
Local Deployment Monitoring
---------------------------
[thumbnail="screenshot/gui-sdn-local-status.png"]
After applying the configuration through the main SDN web-interface panel,
the local network configuration is generated locally on each node in
`/etc/network/interfaces.d/sdn`, and with ifupdown2 reloaded.
After apply configuration on the main sdn section,
the local configuration is generated locally on each node,
in /etc/network/interfaces.d/sdn, and reloaded.
You can monitor the status of local zones && vnets through the main tree.
You can monitor the status of local zones and vnets through the main tree.
Vlan setup example
VLAN Setup Example
------------------
node1: /etc/network/interfaces
TIP: While we show plain configuration content here, almost everything should
be configurable using the web-interface only.
Node1: /etc/network/interfaces
----
auto vmbr0
iface vmbr0 inet manual
@ -252,10 +270,9 @@ iface vmbr0.100 inet static
address 192.168.0.1/24
source /etc/network/interfaces.d/*
----
node2: /etc/network/interfaces
Node2: /etc/network/interfaces
----
auto vmbr0
@ -274,14 +291,15 @@ iface vmbr0.100 inet static
source /etc/network/interfaces.d/*
----
create an vlan zone
Create a VLAN zone named `myvlanzone':
----
id: mylanzone
id: myvlanzone
bridge: vmbr0
----
create a vnet1 with vlan-id 10
Create a VNet named `myvnet1' with `vlan-id` `10' and the previously created
`myvlanzone' as it's zone.
----
id: myvnet1
@ -289,11 +307,12 @@ zone: myvlanzone
tag: 10
----
Apply the configuration on the main sdn section, to create vnets locally on each nodes,
and generate frr config.
Apply the configuration through the main SDN panel, to create VNets locally on
each nodes.
Create a Debian-based Virtual Machine (vm1) on node1, with a vNIC on `myvnet1'.
create a vm1, with 1 nic on vnet1 on node1
Use the following network configuration for this VM:
----
auto eth0
@ -301,19 +320,28 @@ iface eth0 inet static
address 10.0.3.100/24
----
create a vm2, with 1 nic on vnet1 on node2
Create a second Virtual Machine (vm2) on node2, with a vNIC on the same VNet
`myvnet1' as vm1.
Use the following network configuration for this VM:
----
auto eth0
iface eth0 inet static
address 10.0.3.101/24
----
Then, you should be able to ping between between vm1 && vm2
Then, you should be able to ping between both VMs over that network.
QinQ setup example
------------------
node1: /etc/network/interfaces
TIP: While we show plain configuration content here, almost everything should
be configurable using the web-interface only.
Node1: /etc/network/interfaces
----
auto vmbr0
iface vmbr0 inet manual
@ -331,7 +359,7 @@ iface vmbr0.100 inet static
source /etc/network/interfaces.d/*
----
node2: /etc/network/interfaces
Node2: /etc/network/interfaces
----
auto vmbr0
@ -350,7 +378,7 @@ iface vmbr0.100 inet static
source /etc/network/interfaces.d/*
----
create an qinq zone1 with service vlan 20
Create an QinQ zone named `qinqzone1' with service VLAN 20
----
id: qinqzone1
@ -358,7 +386,7 @@ bridge: vmbr0
service vlan: 20
----
create an qinq zone2 with service vlan 30
Create another QinQ zone named `qinqzone2' with service VLAN 30
----
id: qinqzone2
@ -366,7 +394,8 @@ bridge: vmbr0
service vlan: 30
----
create a vnet1 with customer vlan-id 100 on qinqzone1
Create a VNet named `myvnet1' with customer vlan-id 100 on the previously
created `qinqzone1' zone.
----
id: myvnet1
@ -374,7 +403,8 @@ zone: qinqzone1
tag: 100
----
create a vnet2 with customer vlan-id 100 on qinqzone2
Create a `myvnet2' with customer VLAN-id 100 on the previously created
`qinqzone2' zone.
----
id: myvnet2
@ -382,11 +412,12 @@ zone: qinqzone1
tag: 100
----
Apply the configuration on the main sdn section, to create vnets locally on each nodes,
and generate frr config.
Apply the configuration on the main SDN web-interface panel to create VNets
locally on each nodes.
Create a Debian-based Virtual Machine (vm1) on node1, with a vNIC on `myvnet1'.
create a vm1, with 1 nic on vnet1 on node1
Use the following network configuration for this VM:
----
auto eth0
@ -394,14 +425,21 @@ iface eth0 inet static
address 10.0.3.100/24
----
create a vm2, with 1 nic on vnet1 on node2
Create a second Virtual Machine (vm2) on node2, with a vNIC on the same VNet
`myvnet1' as vm1.
Use the following network configuration for this VM:
----
auto eth0
iface eth0 inet static
address 10.0.3.101/24
----
create a vm3, with 1 nic on vnet2 on node1
Create a third Virtual Machine (vm3) on node1, with a vNIC on the other VNet
`myvnet2'.
Use the following network configuration for this VM:
----
auto eth0
@ -409,23 +447,27 @@ iface eth0 inet static
address 10.0.3.102/24
----
create a vm4, with 1 nic on vnet2 on node2
Create another Virtual Machine (vm4) on node2, with a vNIC on the same VNet
`myvnet2' as vm3.
Use the following network configuration for this VM:
----
auto eth0
iface eth0 inet static
address 10.0.3.103/24
----
Then, you should be able to ping between between vm1 && vm2
vm3 && vm4 could ping together
but vm1 && vm2 couldn't ping vm3 && vm4,
as it's a different zone, with different service vlan
Then, you should be able to ping between the VMs 'vm1' and 'vm2', also
between 'vm3' and 'vm4'. But, none of VMs 'vm1' or 'vm2' can ping the VMs 'vm3'
or 'vm4', as they are on a different zone with different service-vlan.
Vxlan setup example
VXLAN Setup Example
-------------------
node1: /etc/network/interfaces
----
auto vmbr0
iface vmbr0 inet static
@ -469,7 +511,9 @@ iface vmbr0 inet static
source /etc/network/interfaces.d/*
----
create an vxlan zone
Create an VXLAN zone named `myvxlanzone', use the lower MTU to ensure the extra
50 bytes of the VXLAN header can fit. Add all previously configured IPs from
the nodes as peer address list.
----
id: myvxlanzone
@ -477,7 +521,8 @@ peers address list: 192.168.0.1,192.168.0.2,192.168.0.3
mtu: 1450
----
create first vnet
Create a VNet named `myvnet1' using the VXLAN zone `myvxlanzone' created
previously.
----
id: myvnet1
@ -485,11 +530,12 @@ zone: myvxlanzone
tag: 100000
----
Apply the configuration on the main sdn section, to create vnets locally on each nodes,
and generate frr config.
Apply the configuration on the main SDN web-interface panel to create VNets
locally on each nodes.
Create a Debian-based Virtual Machine (vm1) on node1, with a vNIC on `myvnet1'.
create a vm1, with 1 nic on vnet1 on node2
Use the following network configuration for this VM, note the lower MTU here.
----
auto eth0
@ -498,7 +544,11 @@ iface eth0 inet static
mtu 1450
----
create a vm2, with 1 nic on vnet1 on node3
Create a second Virtual Machine (vm2) on node3, with a vNIC on the same VNet
`myvnet1' as vm1.
Use the following network configuration for this VM:
----
auto eth0
iface eth0 inet static
@ -506,12 +556,13 @@ iface eth0 inet static
mtu 1450
----
Then, you should be able to ping between between vm1 && vm2
Then, you should be able to ping between between 'vm1' and 'vm2'.
EVPN setup example
------------------
node1: /etc/network/interfaces
----
@ -557,7 +608,8 @@ iface vmbr0 inet static
source /etc/network/interfaces.d/*
----
create a evpn controller
Create a EVPN controller, using a private ASN number and above node addreesses
as peers. Define 'node1' and 'node2' as gateway nodes.
----
id: myevpnctl
@ -566,7 +618,8 @@ peers: 192.168.0.1,192.168.0.2,192.168.0.3
gateway nodes: node1,node2
----
create an evpn zone
Create an EVPN zone named `myevpnzone' using the previously created
EVPN-controller.
----
id: myevpnzone
@ -575,7 +628,8 @@ controller: myevpnctl
mtu: 1450
----
create first vnet
Create the first VNet named `myvnet1' using the EVPN zone `myevpnzone', a IPv4
CIDR network and a random MAC address.
----
id: myvnet1
@ -585,7 +639,8 @@ ipv4: 10.0.1.1/24
mac address: 8C:73:B2:7B:F9:60 #random generate mac addres
----
create second vnet
Create the second VNet named `myvnet2' using the same EVPN zone `myevpnzone', a
different IPv4 CIDR network and a different random MAC address than `myvnet1'.
----
id: myvnet2
@ -595,12 +650,13 @@ ipv4: 10.0.2.1/24
mac address: 8C:73:B2:7B:F9:61 #random mac, need to be different on each vnet
----
Apply the configuration on the main sdn section, to create vnets locally on each nodes,
and generate frr config.
Apply the configuration on the main SDN web-interface panel to create VNets
locally on each nodes and generate the FRR config.
Create a Debian-based Virtual Machine (vm1) on node1, with a vNIC on `myvnet1'.
create a vm1, with 1 nic on vnet1 on node2
Use the following network configuration for this VM:
----
auto eth0
@ -610,7 +666,11 @@ iface eth0 inet static
mtu 1450
----
create a vm2, with 1 nic on vnet2 on node3
Create a second Virtual Machine (vm2) on node2, with a vNIC on the other VNet
`myvnet2'.
Use the following network configuration for this VM:
----
auto eth0
iface eth0 inet static
@ -622,12 +682,14 @@ iface eth0 inet static
Then, you should be able to ping vm2 from vm1, and vm1 from vm2.
from vm2 on node3, if you ping an external ip, the packet will go
to the vnet2 gateway, then will be routed to gateway nodes (node1 or node2)
then the packet will be routed to the node1 or node2 default gw.
If you ping an external IP from 'vm2' on the non-gateway 'node3', the packet
will go to the configured 'myvnet2' gateway, then will be routed to gateway
nodes ('node1' or 'node2') and from there it will leave those nodes over the
default gateway configured on node1 or node2.
Of course you need to add reverse routes to 10.0.1.0/24 && 10.0.2.0/24 to node1,node2 on your external gateway.
If you have configured an external bgp router, the bgp-evpn routes (10.0.1.0/24 && 10.0.2.0/24),
will be announced dynamically.
NOTE: Of course you need to add reverse routes for the '10.0.1.0/24' and
'10.0.2.0/24' network to node1, node2 on your external gateway, so that the
public network can reply back.
If you have configured an external BGP router, the BGP-EVPN routes (10.0.1.0/24
and 10.0.2.0/24 in this example), will be announced dynamically.