OSPF across two areas

What you'll build

• Core — a router, the HQ router that anchors area 0
• Border — a router, the router with a leg in each area — the ABR
• Branch — a router, the branch-office router, alone in area 1
• SW-Core — a switch, the switch the HQ hosts plug into
• PC-HQ — a pc, a workstation at headquarters
• WEB1 — a server, the HQ server the branch has to reach
• PC-Branch — a pc, a workstation at the branch office

Step by step

1. 1. Anchor the backbone at headquarters

   Drag a router and a switch onto the canvas, cable them, and give the router the HQ LAN address 10.10.0.1/24. Area 0 is the backbone every other area has to touch, so it is worth starting where the backbone lives rather than at the far end of the network.

   • Cable Core Gi0/0 ↔ SW-Core Gi0/1

   On Core — Name the HQ router and address its LAN port

   enable
   configure terminal
   hostname Core
   interface Gi0/0
   ip address 10.10.0.1 255.255.255.0
   no shutdown
   exit
   end

   On SW-Core — Name the HQ switch so its console prompt is unmistakable

   enable
   configure terminal
   hostname SW-Core
   end

   Check: run show ip interface brief on Core and look for Gi0/0 10.10.0.1 YES manual up.

   Why: OSPF's areas form a two-level hierarchy with area 0 at the centre, and routes between two areas only ever travel through it. Designing from the backbone outwards means every area you add later has a place to attach from the moment it is built.

2. 2. Put a workstation on the HQ LAN

   Cable a PC to Fa0/1 and give it an address, a mask and the router as its gateway in one line. This host is the yardstick for the rest of the build: every later step is measured by what PC-HQ can and cannot reach.

   • Cable PC-HQ Eth0 ↔ SW-Core Fa0/1

   On PC-HQ — Name the workstation, address it, and prove it reaches its gateway

   hostname PC-HQ
   ipconfig Eth0 10.10.0.20 255.255.255.0 10.10.0.1
   ping 10.10.0.1

   Check: run ip -br a on PC-HQ and look for Eth0 UP 10.10.0.20/24.

   Why: A successful ping to the gateway is a compact proof of several layers at once: the cable and the switch port carried frames, ARP resolved the router's MAC, and both ends agree on the subnet. When a later ping fails, you already know which of those facts it is not.

3. 3. Raise the backbone: area 0 between two routers

   Add the second router, join the two with a /30 — two usable addresses, which is all a point-to-point link ever needs — and start an OSPF process on each. Every `network` statement here says `area 0`, so both routers hold the same link-state database and both run SPF over the same map. That is exactly the property that stops scaling: in one flat area, a flapping link anywhere makes every router everywhere recompute.

   • Cable Core Gi0/1 ↔ Border Gi0/1

   On Core — Address the backbone link and advertise both HQ networks into area 0

   enable
   configure terminal
   interface Gi0/1
   ip address 10.0.0.1 255.255.255.252
   no shutdown
   exit
   router ospf 1
   router-id 1.1.1.1
   network 10.10.0.0 0.0.0.255 area 0
   network 10.0.0.0 0.0.0.3 area 0
   exit
   end

   On Border — Name the border router and bring its area 0 leg into the same process

   enable
   configure terminal
   hostname Border
   interface Gi0/1
   ip address 10.0.0.2 255.255.255.252
   no shutdown
   exit
   router ospf 1
   router-id 2.2.2.2
   network 10.0.0.0 0.0.0.3 area 0
   exit
   end

   Check: run show ip ospf neighbor on Core and look for 2.2.2.2 1 FULL.

   Why: A `network` statement does two jobs at once: it switches OSPF on for every interface whose address it matches, and it places that interface in an area. Area membership is therefore a property of each interface, not of the router — which is exactly what will let one router sit in two areas later in this build.

4. 4. Cable the branch office

   Hang a third router off Border's serial port, give the branch its own /24, and put a workstation on it. Nothing here mentions OSPF yet: this step is only wire and addresses, so that the next one can isolate what the protocol adds.

   • Cable Border Se0/0/0 ↔ Branch Se0/0/0 (serial)
   • Cable Branch Gi0/0 ↔ PC-Branch Eth0

   On Border — Address the serial link towards the branch

   enable
   configure terminal
   interface Se0/0/0
   ip address 10.0.0.5 255.255.255.252
   no shutdown
   exit
   end

   On Branch — Name the branch router and address both of its ports

   enable
   configure terminal
   hostname Branch
   interface Gi0/0
   ip address 10.20.0.1 255.255.255.0
   no shutdown
   exit
   interface Se0/0/0
   ip address 10.0.0.6 255.255.255.252
   no shutdown
   exit
   end

   On PC-Branch — Name and address the branch workstation

   hostname PC-Branch
   ipconfig Eth0 10.20.0.20 255.255.255.0 10.20.0.1

   Check: run show ip route on Branch and look for C 10.0.0.4/30 is directly connected, Se0/0/0.

   Why: Mixing prefix lengths is ordinary: LANs get a /24 because they hold hosts, and each point-to-point link gets a /30 because it holds exactly two routers. OSPF carries every prefix with its own mask, which is what makes a variable-length plan like this one work.

5. 5. Start OSPF at the branch — and watch nothing happen

   Put both branch networks into area 1 and give the router an id. The serial link is up, the addresses are in the same /30, the process is running at this end — and the ping to HQ still fails. Read Branch's neighbour table before you read the next step: the router tells you exactly what it is waiting for.

   On Branch — Advertise the branch LAN and the serial link into area 1

   enable
   configure terminal
   router ospf 1
   router-id 3.3.3.3
   network 10.20.0.0 0.0.0.255 area 1
   network 10.0.0.4 0.0.0.3 area 1
   exit
   end

   On PC-Branch — Try HQ from the branch — this one is supposed to fail

   ping 10.10.0.20

   Check: run show ip ospf neighbor on Branch and look for Confirm the peer also runs OSPF with matching area..

   Why: An OSPF adjacency needs Hellos from both ends of a link, and Border's serial interface matches none of its `network` statements, so it sends none. Branch's Hellos go unanswered, no databases are exchanged, and area 1 stays cut off from the backbone.

6. 6. Give Border a leg in both areas

   One line fixes it: Border claims the serial /30 for area 1 while it keeps Gi0/1 in area 0. A router with interfaces in two areas is an Area Border Router, and an ABR is the only thing that moves reachability between them — it keeps a separate link-state database per area and passes each one a summary of the other instead of the raw topology. Area 1 now has a way into the backbone, and every prefix that crosses arrives marked `O IA`, inter-area.

   On Border — Attach area 1 to the backbone through the existing OSPF process

   enable
   configure terminal
   router ospf 1
   network 10.0.0.4 0.0.0.3 area 1
   exit
   end

   Check: run show ip route on Core and look for O IA 10.20.0.0/24 [110/66] via 10.0.0.2, Gi0/1.

   Why: Between areas, OSPF behaves much like a distance-vector protocol: routers in area 1 know the HQ networks only as reachable through Border at a given cost, never as part of their own map. That is why every area must touch area 0 — with the backbone as the only transit between areas, those second-hand routes cannot form a loop.

7. 7. Add a server to area 0 and read the branch's table

   Plug a server into the HQ switch, address it, and ping it from the branch. Nobody touched Branch, and `show ip route ospf` there looks identical to before — the single `O IA 10.10.0.0/24` entry already covered the new host. That is what buying an area gets you: work inside one area stops at the ABR instead of rippling out as a database change every router in the company has to process.

   • Cable WEB1 Eth0 ↔ SW-Core Fa0/2

   On WEB1 — Name and address the HQ server on the existing LAN

   hostname WEB1
   ipconfig Eth0 10.10.0.80 255.255.255.0 10.10.0.1
   ping 10.20.0.20

   Check: run show ip route ospf on Branch and look for O IA 10.10.0.0/24 [110/66] via 10.0.0.5, Se0/0/0.

   Why: Routing tables list networks, not hosts. WEB1 lives inside 10.10.0.0/24, a prefix Branch already reaches, so plugging it in changes no route anywhere — a new host is never an OSPF event, in any design. What an area boundary hides is topology: changes to links inside area 0 reach Branch only as updated summaries from Border, never as HQ's detailed map.

The theory behind it

• OSPF in a single area

More in Routing

• Static routes across a WAN — Join three sites with point-to-point WAN links and route between them by hand, one line at a time.
• RIP: routing by counting hops — Chain three routers over two serial links and let RIP version 2 teach every router the way to every network, counting the hops as it goes.
• Two routers that learn the network — Join two sites over a WAN link, then let OSPF fill in the routing tables that you would otherwise type by hand.
• A backup link that waits its turn — Run two sites over a fast primary link, park a floating static route on a slow standby line, then pull the primary and watch the backup carry the traffic.
• Steer OSPF with link cost — Give OSPF two paths between two sites, watch it choose the cheaper one over the shorter one, then re-rate a link and watch the route move.
• Share one default route with OSPF — Give the edge router a default route to the provider, then let OSPF hand it to the rest of the campus instead of typing it on every router.