 |
Note: In this guide, the term OSPF refers to OSPF version 2 and OSPF version 3. |
In an Open Shortest Path First (OSPF) network, the network topology is distributed among the nodes of the autonomous system (AS) and is regularly updated through the exchange of link-state advertisements (LSAs). As a result, OSPF is known as a link-state protocol. Because topology changes are flooded throughout the network, every node maintains the same copy of the network map in its local topological database. Packets are then routed based on the shared topology using the shortest path first (SPF) algorithm.
This overview contains the following topics:
|
Note: In this guide, the term OSPF refers to OSPF version 2 and OSPF version 3. |
OSPF creates a topology map by flooding link-state advertisements (LSAs) across OSPF-enabled links. LSAs announce the presence of OSPF-enabled interfaces to adjacent OSPF interfaces. The exchange of LSAs establishes bidirectional connectivity between all adjacent OSPF interfaces (neighbors) using a three-way handshake, as shown in Figure 55.
Figure 55: OSPF Three-Way Handshake
In Figure 55, Router A sends hello packets out all its OSPF-enabled interfaces when it comes online. Router B receives the packet, which establishes that Router B can receive traffic from Router A. Router B generates a response to Router A to acknowledge receipt of the hello packet. When Router A receives the response, it establishes that Router B can receive traffic from Router A. Router A then generates a final response packet to inform Router B that Router A can receive traffic from Router B. This three-way handshake ensures bidirectional connectivity.
As new neighbors are added to the network or existing neighbors lose connectivity, the adjacencies in the topology map are modified accordingly through the exchange (or absence) of LSAs. These LSAs advertise only the incremental changes in the network, which helps minimize the amount of OSPF traffic on the network. The adjacencies are shared and used to create the network topology in the topological database.
Large local area networks (LANs) that have many routers and therefore many OSPF adjacencies can produce heavy control-packet traffic as LSAs are flooded across the network. To alleviate the potential traffic problem, OSPF uses designated routers (DRs). Rather than broadcasting LSAs to all their OSPF neighbors, the routers send their LSAs to the designated router, which processes the LSAs, generates responses, and multicasts topology updates to all OSPF routers.
In LANs, the election of the designated router takes place when the OSPF network is initially established. When the first OSPF links are active, the router with the highest router identifier (defined by the router-id configuration value or the loopback address) is elected designated router. The router with the second highest router identifier is elected the backup designated router (BDR). If the designated router fails or loses connectivity, the BDR assumes its role and a new BDR election takes place between all the routers in the OSPF network.
Once the topology is shared across the network, OSPF uses it to route packets between network nodes. Each path between neighbors is assigned a cost based on the throughput, round-trip time, and reliability of the link. The sum of the costs across a particular path between hosts determines the overall cost of the path. Packets are then routed along the shortest path using the shortest path first (SPF) algorithm. If multiple equal-cost paths exist between a source and destination address, OSPF routes packets along each path alternately, in round-robin fashion.
OSPF allows you to manually assign a cost (or metric) to a particular path segment to control the flow of packets across the network.
The OSPF networks in an AS are administratively grouped into areas. Each area within an AS operates like an independent network and has a unique 32-bit area ID, which functions like a network address. Within an area, the topology database contains only information about the area, LSAs are flooded only to nodes within the area, and routes are computed only within the area. Subnetworks are divided into other areas, which are connected to form the whole of the main network.
The central area of an AS, called the backbone area, has a special function and is always assigned the area ID 0.0.0.0. Area IDs are unique numeric identifiers, in dotted decimal notation, but they are not IP addresses. Area IDs need only be unique within an AS. All other networks or areas in the AS must be directly connected to the backbone area by a router that has interfaces in more than one area. These connecting routers are called area border routers (ABRs). Figure 56 shows an OSPF topology of three areas connected by two area border routers.
Figure 56: Multiarea OSPF Topology
Area border routers are responsible for sharing topology information between areas. They summarize the link-state records of each area and advertise destination address summaries to neighboring areas. The advertisements contain the ID of the area in which each destination lies, so that packets are routed to the appropriate area border router. For example, in the OSPF areas shown in Figure 56, packets sent from Router A to Router C are automatically routed through Area Border Router B.
An OSPF restriction requires all areas to be directly connected to the backbone area so that packets can be properly routed. All packets are routed first to the backbone area by default. Packets that are destined for an area other than the backbone area are then routed to the appropriate area border router and on to the remote host within the destination area.
In large networks with many areas, in which direct connectivity between all areas and the backbone area is physically difficult or impossible, you can configure virtual links to connect noncontiguous areas. For example, Figure 57 shows a virtual link between a noncontiguous area and the backbone area through an area connected to both.
Figure 57: OSPF Topology with a Virtual Link
In the topology shown in Figure 57, a virtual link is established between area 0.0.0.3 and the backbone area through area 0.0.0.2. All outbound traffic destined for other areas is routed through area 0.0.0.2 to the backbone area and then to the appropriate area border router. All inbound traffic destined for area 0.0.0.3 is routed to the backbone area and then through area 0.0.0.2.
Figure 58 shows an AS across which many external routes are advertised. If external routes make up a significant portion of a topology database, you can suppress the advertisements in areas that do not have links outside the network. By doing so, you can reduce the amount of memory the nodes use to maintain the topology database and free it for other uses.
Figure 58: OSPF AS Network with Stub Areas and NSSAs
To control the advertisement of external routes into an area, OSPF uses stub areas. By designating an area border router interface to the area as a stub interface, you suppress external route advertisements through the area border router. Instead, the area border router automatically advertises a default route (through itself) in place of the external routes. Packets destined for external routes are automatically sent to the area border router, which acts as a gateway for outbound traffic and routes them appropriately.
For example, area 0.0.0.3 in Figure 58 is not directly connected to the outside network. All outbound traffic is routed through the area border router to the backbone and then to the destination addresses. By designating area 0.0.0.3 a stub area, you reduce the size of the topology database for that area by limiting the route entries to only those routes internal to the area.
Like area 0.0.0.3 in Figure 58, area 0.0.0.4 has no external connections. However, area 0.0.0.4 has static customer routes that are not internal OSPF routes. You can limit the external route advertisements to the area and advertise the static customer routes by designating it a not-so-stubby area (NSSA). External routes are flooded into the NSSA and then leaked to the other areas, but external routes from other areas are not advertised within the NSSA.
