Problems with large-scale OSPF network

OSPF is a type of IGPs that is used within an AS (Autonomous System), and let’s imagine that the OSPF network within the AS becomes larger and larger, for example, with dozens to hundreds of routers. In such a very large network, the following problems can be considered.

  • The number of LSAs flowing on the network will increase, overwhelming the network bandwidth
  • The link state database gets larger
  • Routing table gets larger
  • Increased impact of changes in the network
  • Increased memory and CPU load on routers

The number of LSAs flowing on the network will increase, overwhelming the network bandwidth

In OSPF, each router exchanges link state advertisement(LSA). Of course, the more routers and the more networks a router has, the more LSAs it will have. Then, the exchange of LSAs will overwhelm the bandwidth of the network considerably.

The link state database gets larger

The increase in the number of LSAs not only overwhelms the bandwidth of the network, but also increases the size of the link state database maintained by each router.

Routing table gets larger

Furthermore, since the routing table is calculated from the link state database according to the SPF algorithm, the size of the routing table will inevitably increase as the size of the link state database increases. Not only that, but it also increases the time required for the SPF algorithm to calculate the routing table.

Increased impact of changes in the network

When there is a change in the network, the router that detects the change must send the change information to other OSPF routers to keep the link state database synchronized across the network. Therefore, a change in one part of the network can affect the entire network. If a situation arises where a certain part of a large network repeatedly goes up and down, each time the LSA is flooded, it will result in a situation where the SPF calculation in the routing table from the link state database will be frequent.

Increased memory and CPU load on routers

SPF calculation places a considerable load on the router’s CPU, so it may become overwhelmed with SPF calculation alone and be unable to perform other processes. A large network consists of a number of sites. In most cases, a network outage in one site has no effect on communications in other sites. However, in a single area, a downed network in one site may cause SPF calculations to be performed on OSPF routers in all sites. During this time, no packets can be routed.

Figure Problems with large-scale OSPF network

OSPF area division

In order to solve the various problems that can occur in such a large OSPF network, a concept called “area” is very important. By dividing the OSPF network into appropriate areas, OSPF can support large networks.

By dividing the area, LSDB synchronization is maintained with OSPF routers in the same area. And the OSPF router has a detailed understanding of the network diagram in LSDB. By dividing the area, there is a different level of detail to understand the network diagram inside and outside the area. Inside the area, the network diagram will be very detailed. On the other hand, outside the area, only the network address will be known, not the detailed network diagram.

LSAs that describe details in an area are advertised only in the area. This reduces the LSA traffic to be exchanged and also reduces the size of the LSDB and routing table. Then, for changes outside the area, the SPF will not be calculated to reduce the impact of changes in less relevant areas.

How the OSPF works