[neutron][victoria] Front / Back Routers
All; My team and I have been working through the tutorials on Server World (server-world.com/ean/), in order to learn and build an OpenStack cluster. We've also been looking at the official documentation to attempt to increase our knowledge of the subject. I have a question about Neutron though. All the examples that I remember have Neutron setup with a single router. The router is part of a "provider" network, and subnet on the outside, and one or more "tenant" networks on the inside. Floating IPS, then appear to be IP addresses belonging to the "provider" subnet, that are applied to the router, and which the router then NATs. These setups look like this: Physrouter1 (physical router) subnet: 192.168.0.0/24, IP address: 192.168.0.1 | Physnet1 (192.168.0.0/24)(ovs network definition) | Router1 (ovs router)(allocation pool: 192.168.0.100 - 192.168.0.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s) This has the advantage of being easy, fast, secure, and simple to setup. What if you wanted something where you could route whole subnet into your OpenStack cluster. Physrouter1 (physical router) subnet: 172.16.255.0/24, IP address: 172.16.255.1 | Physnet1 (172.16.255.0/24)(ovs network definition) | Router1 (ovs router)(fixed IP addresses: 172.16.255.2 & 172.16.254.1/24 + static / dynamic routing) | Network (17216.254.0/24) | Router2(ovs router)(allocation pool: 172.16.254.5 - 172.16.254.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s) Is my understanding accurate? Are there advantages of one over the other? What commands are used to specify static IPs for ovs routers, and static routing rules? Thank you, Dominic L. Hilsbos, MBA Director - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com
It's a bit hard to parse a network topology by email, but from a theoretical point of view - you can statically route a /24 towards the external IP of an openstack router from a device upstream. I do believe there is a BGP component for Openstack but I'm not sure it's role is to dynamically advertised networks from Openstack towards the wider network. On Thu, Feb 4, 2021 at 11:45 AM <DHilsbos@performair.com> wrote:
All;
My team and I have been working through the tutorials on Server World ( server-world.com/ean/), in order to learn and build an OpenStack cluster. We've also been looking at the official documentation to attempt to increase our knowledge of the subject.
I have a question about Neutron though. All the examples that I remember have Neutron setup with a single router. The router is part of a "provider" network, and subnet on the outside, and one or more "tenant" networks on the inside. Floating IPS, then appear to be IP addresses belonging to the "provider" subnet, that are applied to the router, and which the router then NATs.
These setups look like this:
Physrouter1 (physical router) subnet: 192.168.0.0/24, IP address: 192.168.0.1 | Physnet1 (192.168.0.0/24)(ovs network definition) | Router1 (ovs router)(allocation pool: 192.168.0.100 - 192.168.0.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s)
This has the advantage of being easy, fast, secure, and simple to setup.
What if you wanted something where you could route whole subnet into your OpenStack cluster.
Physrouter1 (physical router) subnet: 172.16.255.0/24, IP address: 172.16.255.1 | Physnet1 (172.16.255.0/24)(ovs network definition) | Router1 (ovs router)(fixed IP addresses: 172.16.255.2 & 172.16.254.1/24 + static / dynamic routing) | Network (17216.254.0/24) | Router2(ovs router)(allocation pool: 172.16.254.5 - 172.16.254.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s)
Is my understanding accurate? Are there advantages of one over the other? What commands are used to specify static IPs for ovs routers, and static routing rules?
Thank you,
Dominic L. Hilsbos, MBA Director - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com
Hi, On Thu, Feb 04, 2021 at 10:08:01PM -0500, Laurent Dumont wrote:
It's a bit hard to parse a network topology by email, but from a theoretical point of view - you can statically route a /24 towards the external IP of an openstack router from a device upstream.
I do believe there is a BGP component for Openstack but I'm not sure it's role is to dynamically advertised networks from Openstack towards the wider network.
It is neutron-dynamic-routing. Documentation is available at https://docs.openstack.org/neutron-dynamic-routing/victoria/
On Thu, Feb 4, 2021 at 11:45 AM <DHilsbos@performair.com> wrote:
All;
My team and I have been working through the tutorials on Server World ( server-world.com/ean/), in order to learn and build an OpenStack cluster. We've also been looking at the official documentation to attempt to increase our knowledge of the subject.
I have a question about Neutron though. All the examples that I remember have Neutron setup with a single router. The router is part of a "provider" network, and subnet on the outside, and one or more "tenant" networks on the inside. Floating IPS, then appear to be IP addresses belonging to the "provider" subnet, that are applied to the router, and which the router then NATs.
These setups look like this:
Physrouter1 (physical router) subnet: 192.168.0.0/24, IP address: 192.168.0.1 | Physnet1 (192.168.0.0/24)(ovs network definition) | Router1 (ovs router)(allocation pool: 192.168.0.100 - 192.168.0.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s)
This has the advantage of being easy, fast, secure, and simple to setup.
What if you wanted something where you could route whole subnet into your OpenStack cluster.
Physrouter1 (physical router) subnet: 172.16.255.0/24, IP address: 172.16.255.1 | Physnet1 (172.16.255.0/24)(ovs network definition) | Router1 (ovs router)(fixed IP addresses: 172.16.255.2 & 172.16.254.1/24 + static / dynamic routing) | Network (17216.254.0/24) | Router2(ovs router)(allocation pool: 172.16.254.5 - 172.16.254.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s)
Is my understanding accurate? Are there advantages of one over the other? What commands are used to specify static IPs for ovs routers, and static routing rules?
Thank you,
Dominic L. Hilsbos, MBA Director - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com
-- Slawek Kaplonski Principal Software Engineer Red Hat
Hi, On Thu, Feb 04, 2021 at 04:39:23PM +0000, DHilsbos@performair.com wrote:
All;
My team and I have been working through the tutorials on Server World (server-world.com/ean/), in order to learn and build an OpenStack cluster. We've also been looking at the official documentation to attempt to increase our knowledge of the subject.
I have a question about Neutron though. All the examples that I remember have Neutron setup with a single router. The router is part of a "provider" network, and subnet on the outside, and one or more "tenant" networks on the inside. Floating IPS, then appear to be IP addresses belonging to the "provider" subnet, that are applied to the router, and which the router then NATs.
These setups look like this:
Physrouter1 (physical router) subnet: 192.168.0.0/24, IP address: 192.168.0.1 | Physnet1 (192.168.0.0/24)(ovs network definition)
Can You explain more what "ovs network definition" means really?
| Router1 (ovs router)(allocation pool: 192.168.0.100 - 192.168.0.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool
What is "ovs router" here? Can You explain? In Neutron router don't have allocation pool. Allocation pool is attribute of the subnet defined in the network. Subnet can be then plugged to the router. Or if it's "router:external" network, it can be used as a gateway for the router and then You can have Floating IPs from it.
| Tenant network(s)
This has the advantage of being easy, fast, secure, and simple to setup.
What if you wanted something where you could route whole subnet into your OpenStack cluster.
IIUC You can route /24 on Your physical router and then configure it as subnet in the provider network (vlan or flat). If You will set it as router:external, You will be able to use it as neutron router's gateway and have FIPs from it. But You can also plug vms directly to that network. Neutron will then allocation IP addresses from Your /24 subnet to the instances.
Physrouter1 (physical router) subnet: 172.16.255.0/24, IP address: 172.16.255.1 | Physnet1 (172.16.255.0/24)(ovs network definition) | Router1 (ovs router)(fixed IP addresses: 172.16.255.2 & 172.16.254.1/24 + static / dynamic routing) | Network (17216.254.0/24) | Router2(ovs router)(allocation pool: 172.16.254.5 - 172.16.254.254) <-- Floating IPs are "owned" by this, and are in the range of the allocation pool | Tenant network(s)
Is my understanding accurate? Are there advantages of one over the other? What commands are used to specify static IPs for ovs routers, and static routing rules?
Thank you,
Dominic L. Hilsbos, MBA Director - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com
-- Slawek Kaplonski Principal Software Engineer Red Hat
participants (3)
-
DHilsbos@performair.com
-
Laurent Dumont
-
Slawek Kaplonski