Bandwidth for IP interfaces

The bandwidth for an IP interface can be specified so that higher level protocols, such as OSPFv2 and OSPFv3, can use this setting to influence the routing cost for routes learned on these interfaces.

When the interface bandwidth is configured, the number of network and router link state advertisement generation is reduced during an operation down or a shutdown of one or more of the associated interfaces of the VE interface. For OSPF, when the dynamic cost feature is enabled, the bandwidth for a VE interface is the sum of bandwidth for either all associated ports or all active associated ports. However, when the interface bandwidth is configured on the VE interface itself, the bandwidth of the associated ports are not used in the OSPF cost calculation. This means that even when one of the associated ports of the VE interface goes down, there is no OSPF cost recalculation.

The bandwidth for IP interfaces feature can be configured for a physical interface, Link aggregation (LAG) groups, a VE interface, and a tunnel interface.

The bandwidth for IP interfaces feature can be used to:

  • Query the bandwidth for an interface.
  • Help OSPF avoid generating numerous LSAs while updating the cost value for a VE interface due to changes in associated physical interfaces.
  • Influence the cost on OSPF interfaces for specific tunnels, VE interfaces, and physical interfaces.

The bandwidth for IP interfaces feature enables OSPF to calculate its interface metric cost more precisely, based on the specified interface bandwidth. If the interface bandwidth feature is disabled, OSPF calculates the cost as the reference-bandwidth divided by the fixed port bandwidth, as outlined in the Changing the reference bandwidth for the cost on OSPF interfaces section. When the interface bandwidth feature is enabled, OSPF calculates the cost as the reference-bandwidth divided by the interface bandwidth. For a physical interface, the interface bandwidth is assigned by default to the port speed.

The interface bandwidth feature also enables OSPF to use the configured interface bandwidth for a VE interface to calculate its routing metric, without considering the bandwidth of the associated physical ports. When this feature is enabled, the bandwidth for a VE interface is the interface bandwidth value if it is configured under the VE. Alternatively, it is the sum of the interface bandwidth for all associated ports or all active ports when OSPF dynamic cost is enabled.

The bandwidth of a trunk port for OSPF is, by default, the sum of either all the associated ports or all active associated ports when OSPF dynamic cost is enabled. The interface bandwidth of the primary port is used if the interface bandwidth is configured; otherwise it reverts to the default behavior.

NOTE
If the interface bandwidth configuration of the primary port is different to any of the secondary ports, then the LAG is not deployed. When the LAG is undeployed, the interface bandwidth value for all secondary ports is reset to the port speed.

The configured value is exposed in SNMP via ifSpeed (in ifTable) and ifHighSpeed (in ifXTable) objects.

NOTE
GRE or IPv6 tunnel bandwidth may limit routing protocol traffic propagating through the tunnel. For example, if the tunnel defaults to 8kbps , OSPF uses 50% of the tunnel bandwidth for Hello and update traffic. Therefore, it is good practice to increase the tunnel bandwidth when a routing protocol runs over it to eliminate flapping, and give the routing protocol more capacity to send its update and Hello messages.

From FastIron Release 08.0.30, this feature is supported on all platforms.

Limitations and pre-requisites

  • The bandwidth for IP interfaces feature does not support setting and adjusting GRE or IPv6 receiving and transmission bandwidth.
  • SNMP does not support any IP interface bandwidth related configurations.