Connecting Multiple OSPF Areas

An OSPF area is defined as a logical grouping of routers by a network administrator. OSPF routers in any area share the same topological view (also known as the OSPF database) of the network. The core reason that OSPF is configured in multiple areas is to reduce routing table sizes, which in turn reduces the topological database and CPU/memory requirements on a router.

OSPF is not just configured in one large area, so all routers share the same topological database. The use of multiple areas ensures that the flooding and database management required in large OSPF networks is reduced within each area so that the process of flooding the full database and maintaining full network connectivity does not consume a large portion of the CPU processing power. Every time a network change occurs, the CPU on a router is interrupted and a new OSPF tree is calculated. Running the shortest path first (SPF) algorithm itself is not CPU intensive, but sending and flooding the network with new topological information is extremely CPU intensive.

Routing tables become very large even with only 50 routers. The OSPF database is exchanged every 30 minutes in full, and if this database is too large, every time the exchange occurs, the amount of bandwidth used over the network increases, which can cause severe delays in sending user-based traffic because convergence times increase.

Considering the demands on CPU and memory along with reduced IP routing tables, you should now have a good understanding of why OSPF requires more than one area. In Scenario 3-2 in Chapter 3, you saw how to configure an OSPF network that is partitioned from the backbone. All OSPF areas must be connected to the backbone in case of network failure. When an area cannot reside physically or logically on the backbone, a virtual link is required. For partitioned areas, OSPF treats the area as a separate area, and no routing information flows to the backbone; therefore, you do not have IP connectivity.

Virtual links add a layer of complexity and might cause additional problems when applied to large IP networks. It is best to avoid virtual links in the real world.

When configuring a virtual link, you must be aware of the following design restrictions:

• Virtual links must be configured between two area border routers (ABRs).

• The transit area cannot be a stub area.

• The transit area must have full routing knowledge of both partitioned areas. NOTE

Stub areas are covered later in this chapter. Remember that all routers must be connected to the backbone logically or you must use a virtual link. To understand why logical links are required in today's networks, consider the case were Company XYZ buys Company ACME. Both companies use OSPF and have their own individual backbones. Rather than re-address the networks, a virtual link can provide immediate IP connectivity.

Table 4-1 summarizes the four OSPF area types and their functions.

Table 4-1. OSPF Router Types

Router Type

Description

Internal router

This router is within a specific area only. Internal router functions include maintaining the OSPF database and forwarding data to other networks. All interfaces on internal routers are in the same area.

Area border router (ABR)

ABRs are responsible for connecting two or more areas. An ABR contains the full topological database for each area it is connected to and sends this information to other areas.

Autonomous system boundary router (ASBR)

ASBRs connect to the outside world or perform some form of redistribution into OSPF.

Backbone router

Backbone routers are connected to area 0, which is also represented as area 0.0.0.0. Backbone routers can be internal routers or ASBRs.

Figure 4-1 displays a typical OSPF area assignment and the function of these routers.

Figure 4-1 displays a typical OSPF area assignment and the function of these routers.

Figure 4-1. Typical OSPF Area Assignment and OSPF

Routers

Figure 4-1. Typical OSPF Area Assignment and OSPF

Routers

In Figure 4-1, the routers residing in the backbone (area 0) are called backbone routers. A backbone router connecting to another area can also be an ABR. Routers that connect to, for example, the Internet and redistribute external IP routing tables from such protocols as Border Gateway Protocol (BGP) are termed autonomous system boundary routers (ASBRs). So, you can have a backbone router perform ASBR functions as well as ABR functions.

Each router, depending on its function, sends out a link-state advertisement (LSA). An LSA is a packet used by such routing protocols as OSPF (that is, link-state routing protocols) to send information to neighboring routers describing networks and path costs.

Before flooding any neighboring routers with LSAs, Cisco IOS routers must first undergo the following:

Step 1. Ensure the neighboring router is in a state of adjacency.

Step 2. The interface cannot be a stub area (LSA type 5. Stub areas are discussed later in this chapter.)

Step 3. The interface cannot be connected to a totally stubby area. (LSA type 3, 4, or 5 will not be sent. Totally stubby areas are discussed later in this chapter.)

For a detailed summary of OSPF and the packet types, the Cisco Press titles Routing TCP/IP, Volumes I and II, by Jeff Doyle and Jennifer DeHaven Carroll (Volume II only) explain all the advanced concepts you could ever need.

OSPF supports a number of LSA types as well as three other area types: a stub area, a totally stubby area, and a not-so-stubby area (NSSA). These additional areas provide even more functionality in OSPF. Before covering these new areas in detail, this section first goes over the link-state advertisement types and when to use them in an OSPF environment.

The OSPF standard defines a number of LSAs types. Unlike distance vector protocols (for example, RIP), OSPF does not actually send its routing table to other routers. Instead, OSPF sends the LSA database and derives the IP routing table from LSAs. Table 4-2 describes the six most common LSAs and their functions.

Table 4-2. Six Common Supported LSA Types on Cisco IOS Routers

LSA Packet

Type

Name

Function

1

Router link advertisements

Describes the state and cost of the router's own interfaces.

2

Network link advertisements

Used on multiaccess networks. These are originated by the designated router (DR).

3

Summary link advertisements (ABRs)

Originated by ABRs only. This LSA type sends out information into the autonomous system

(AS) but outside of the area (interarea routes).

4

Summary link advertisements (ASBRs)

Originated by ASBRs describing IP networks external to the AS.

5

Autonomous system (AS) external link advertisements

An LSA sent to a router that connects to the Internet, for example. An advertisement sent from ABR to the ASBR.

6

Not-so-stubby areas (NSSA)

An advertisement bound to an NSSA area.

A stub area is defined as an area that contains a single exit point from the area. A stub in the English dictionary means a dead end, and that is exactly what it means in OSPF. Areas that reside on the edge of the network with no exit point except one path can be termed a stub area. Stubs come in three types.

Table 4-3 summarizes the functions of these new areas, called stubby areas, total stubby areas, and not-so-stubby areas. Take important note of the LSA type allowed or not allowed to fully appreciate the value of a stub area.

Table 4-3. Additional Area Types

Area Type

Function

Stub area

This area does not accept LSA types 4 and 5, which are summary links and external link advertisements, respectively. The only way to achieve a route to unknown destinations is, thereby, a default route injected by the ABR.

Totally stubby area

This area blocks LSA types 3, 4, and 5. Although similar to a stub area, a totally stubby area blocks LSAs of type 3 as well. This solution is Cisco-proprietary and is used to further reduce a topological database.

Not-so-

stubby area

This area is used primarily for connections to an ISP. This area is designed to allow LSAs of type 7 only. All advertised routes can be flooded through the NSSA but are blocked by the ABR. Basically, a type 7 LSA (if the P bit is set to one) will be convert to a type 5 LSA and flooded throughout the rest of the network. If the P bit is set to zero, no translation takes place. Type 4 or 5 LSAs are not permitted. This advertisement will not be propagated to the rest of the network. Typically used to provide a default route.

The only way to appreciate these new areas is to configure them and view the OSPF database. The scenarios that follow cover stub, totally stubby, and not-so-stubby areas in more detail.

NOTE

A stub area cannot be a transit for a virtual link. This is a design limitation by the protocol itself. When a router is defined as a stub area, a bit, called the E bit, in the Hello packet is set to 0. All routers that form any OSPF neighbor relationship must have the E bit set to 0 as well; otherwise, no adjacency is formed.

Also a stub (does not permit LSA types 4 and 5) area or totally stubby (does not permit LSA types 3, 4, and 5) area does not allow external routes. Nor is redistribution allowed. Those functions must be performed by ABRs or ASBRs.

Table 4-4 summarizes the LSA types by area and indicates which LSAs are permitted or disallowed in certain areas.

Table 4-4. LSA Types and Area Restrictions

LSA Type Permitted?

Area

1/2

3/4

6

7

NSSA

Yes

Yes

No

Yes

Totally stubby

Yes

No

No

No

Stub

Yes

Yes

No

No

All OSPF packets are sent using IP protocol port number 89. OSPF runs over the IP layer (also called the Network layer) of the Open System Interconnection (OSI) model.

Was this article helpful?

0 0

Post a comment