OSPF on Cisco Catalyst Routers

[fowlrock]

Any CCNA's want to take a stab at this? Looking for best practice.

All sites connected via ELAN (Any/Any QnQ)

Site 1:
Networks: 10.10.1.1/24 (WAN Vlan). LAN Subnets - 10.0.0.0/24, 10.1.0.0/24, 10.1.3.0/24

Site 2:
Networks: 10.10.1.2/24 (WAN Vlan). LAN Subnet - 192.168.0.0/24

Site 3:
Networks: 10.10.1.3/24 (WAN Vlan). LAN Subnet - 192.168.1.0/24

Site 4:
Networks: 10.10.1.4/24 (WAN Vlan). LAN Subnet - 192.168.2.0/24

Sites 2,3 and 4 are all one Company, but separate branches. Full broadcast between them

Site 1 is a DC that only wants those 3 10.X.X.X/24 networks broadcasted to the other 3 sites.

What would be the cleanest configuration to achieve this? An Area at Site 1 with those networks listed? Treat the other 3 branches as a stub area? Broadcast it all and let the ACL's do the summarization?

 

[scrappymouse]

Not sure if it is a type but your 3 10.10.1.1/24, 10.10.1.2/24, and 10.10.1.3/24 are all in the same network, did you mean 10.10.1.1/24, 10.10.2.1/24, and 10.10.3.1/24 networks?

 

[fowlrock]

Not sure if it is a type but your 3 10.10.1.1/24, 10.10.1.2/24, and 10.10.1.3/24 are all in the same network, did you mean 10.10.1.1/24, 10.10.2.1/24, and 10.10.3.1/24 networks?

Nope. So since these 4 sites are connected via ELAN, I'll use a single "WAN" vlan. They each have an IP on this subnet. Think of it as a "management" vlan.

 

[scrappymouse]

Nope. So since these 4 sites are connected via ELAN, I'll use a single "WAN" vlan. They each have an IP on this subnet. Think of it as a "management" vlan.

Understand I think it was just the "networks" keyword at the start.

 

[Daedalus0101101]

ScrappyMouse is right. You can't use the same subnet for the WAN Links, At least not reliably and I've never gotten it to work. By using a routing protocol whether it be OSPF, BGP ,or EIGRP you are essentially advertising to other routers that you own that entire subnet (depending on subnet mask length). So having all of them in the same subnet with a CIDR of 24, all of your routers are essentially trying to tell each other ""I OWN the 10.10.1.0/24 subnet". Another way to look at it is if I were to try and tell everyone on the board that my username is actually fowlrock as well.

As for how to create connectivity between them all, it really goes down to a few main factors. 1) is the bandwidth at the DC Site and 2) the type of router at the DC site. For simplicity, a hub and spoke ( some call it a partial mesh) setup is usually the easiest, but requires more power at the DC site. The Bandwidth needs to be able to handle all the traffic that is going to be directed to the DC and each sub site as well as between each sub site. Meaning that if there is traffic from say Branch 1 that needs to go to Branch 2, it has to go through the DC to get there. The other way is a Full Mesh setup. All routers are talking with each other directly which reduces load and bandwidth on the DC router, but requires more work from the Network Engineer to make it all work nicely. A Full Mesh also requires that depending on your number of locations that each of the branch routers might need to be more powerful to handle the network load and configurations. One thing that was a huge thing that helped with that is something called MPLS ( Multi-Protocol Label Switching) or PNT ( Private Network Transport). It requires only a single connection and usually a fairly simple routing configuration.

 

[fowlrock]

ScrappyMouse is right. You can't use the same subnet for the WAN Links, At least not reliably and I've never gotten it to work. By using a routing protocol whether it be OSPF, BGP ,or EIGRP you are essentially advertising to other routers that you own that entire subnet (depending on subnet mask length). So having all of them in the same subnet with a CIDR of 24, all of your routers are essentially trying to tell each other ""I OWN the 10.10.1.0/24 subnet". Another way to look at it is if I were to try and tell everyone on the board that my username is actually fowlrock as well.

As for how to create connectivity between them all, it really goes down to a few main factors. 1) is the bandwidth at the DC Site and 2) the type of router at the DC site. For simplicity, a hub and spoke ( some call it a partial mesh) setup is usually the easiest, but requires more power at the DC site. The Bandwidth needs to be able to handle all the traffic that is going to be directed to the DC and each sub site as well as between each sub site. Meaning that if there is traffic from say Branch 1 that needs to go to Branch 2, it has to go through the DC to get there. The other way is a Full Mesh setup. All routers are talking with each other directly which reduces load and bandwidth on the DC router, but requires more work from the Network Engineer to make it all work nicely. A Full Mesh also requires that depending on your number of locations that each of the branch routers might need to be more powerful to handle the network load and configurations. One thing that was a huge thing that helped with that is something called MPLS ( Multi-Protocol Label Switching) or PNT ( Private Network Transport). It requires only a single connection and usually a fairly simple routing configuration.

Yeah - you are over thinking this. 10.10.1.0 is a management subnet - so each site will have a different router ID on that subnet. They will all be in the same OSPF area. There will be no "fighting" over the mgmt subnet because they will all see each other as OSPF Neighbors on that spanned Vlan.