A link bundle is simply a group of ports that are bundled together and act as a single link. The advantages of link bundles are these:

• Multiple links can span several line cards to form a single interface. Thus, the failure of a single link does not cause a loss of connectivity.

• Bundled interfaces increase bandwidth availability, because traffic is forwarded over all available members of the bundle. Therefore, traffic can flow on the available links if one of the links within a bundle fails. Bandwidth can be added without interrupting packet flow.

Although the individual links within a single bundle can have varying speeds, all links within a bundle must be of the same type.

Cisco IOS XR software supports these methods of forming bundles of Ethernet interfaces:

• IEEE 802.3ad—Standard technology that employs a Link Aggregation Control Protocol (LACP) to ensure that all the member links in a bundle are compatible. Links that are incompatible or have failed are automatically removed from a bundle.

• EtherChannel—Cisco proprietary technology that allows the user to configure links to join a bundle, but has no mechanisms to check whether the links in a bundle are compatible.

This list describes the properties and limitations of link bundles:

• Any type of Ethernet interfaces can be bundled, with or without the use of LACP (Link Aggregation Control Protocol).

• Bundle membership can span across several line cards that are installed in a single router.

• A single bundle supports maximum of 64 physical links.

• Different link speeds are allowed within a single bundle, with a maximum of four times the speed difference between the members of the bundle.

• Physical layer and link layer configuration are performed on individual member links of a bundle.

• Configuration of network layer protocols and higher layer applications is performed on the bundle itself.

• A bundle can be administratively enabled or disabled.

• Each individual link within a bundle can be administratively enabled or disabled.

• Ethernet link bundles are created in the same way as Ethernet channels, where the user enters the same configuration on both end systems.

• The MAC address that is set on the bundle becomes the MAC address of the links within that bundle.

• When LACP configured, each link within a bundle can be configured to allow different keepalive periods on different members.

• Load balancing (the distribution of data between member links) is done by flow instead of by packet. Data is distributed to a link in proportion to the bandwidth of the link in relation to its bundle.

• QoS is supported and is applied proportionally on each bundle member.

• Link layer protocols, such as CDP and HDLC keepalives, work independently on each link within a bundle.

• Upper layer protocols, such as routing updates and hellos, are sent over any member link of an interface bundle.

• Bundled interfaces are point to point.

• A link must be in the up state before it can be in distributing state in a bundle.

• All links within a single bundle must be configured either to run 802.3ad (LACP) or Etherchannel (non-LACP). Mixed links within a single bundle are not supported.

• A bundle interface can contain physical links and VLAN subinterfaces only.

• Access Control List (ACL) configuration on link bundles is identical to ACL configuration on regular interfaces.

• Multicast traffic is load balanced over the members of a bundle. For a given flow, internal processes select the member link and all traffic for that flow is sent over that member.

The IEEE 802.3ad standard typically defines a method of forming Ethernet link bundles.

For each link configured as bundle member, this information is exchanged between the systems that host each end of the link bundle:

• A globally unique local system identifier

• An identifier (operational key) for the bundle of which the link is a member

• An identifier (port ID) for the link

• The current aggregation status of the link

This information is used to form the link aggregation group identifier (LAG ID). Links that share a common LAG ID can be aggregated. Individual links have unique LAG IDs.

The system identifier distinguishes one router from another, and its uniqueness is guaranteed through the use of a MAC address from the system. The bundle and link identifiers have significance only to the router assigning them, which must guarantee that no two links have the same identifier, and that no two bundles have the same identifier.

The information from the peer system is combined with the information from the local system to determine the compatibility of the links configured to be members of a bundle.

Bundle MAC addresses in the routers come from a set of reserved MAC addresses in the backplane. This MAC address stays with the bundle as long as the bundle interface exists. The bundle uses this MAC address until the user configures a different MAC address. The bundle MAC address is used by all member links when passing bundle traffic. Any unicast or multicast addresses set on the bundle are also set on all the member links.

Note	We recommend that you avoid modifying the MAC address, because changes in the MAC address can affect packet forwarding.

In LACP, by default, a higher priority port would become the active port after it becomes operational again. To avoid this reversion, you can run the lacp non-revertive command. This configures the lower priority port to continue as the active port even after the higher priority port is capable of being operational. This avoids the traffic disruption that may happen in putting the currently active but lower priority port into standby and diverting traffic through the higher priority port that is now capable of being operational.

On the ingress direction, QoS is applied to the local instance of a bundle. Each bundle is associated with a set of queues. QoS is applied to the various network layer protocols that are configured on the bundle.

On the egress direction, QoS is applied on the bundle with a reference to the member links. QoS is applied based on the sum of the member bandwidths.

When QoS is applied on the bundle for either the ingress or egress direction, QoS is applied at each member interface.

The Link Bundling feature supports all the QoS features described in the Cisco ASR 9000 Series Aggregation Services Router Modular Quality of Service Configuration Guide.

The Link Bundling feature supports these QoS features:

• hi priority /lo priority—Maximum bandwidth is calculated as a percentage of the bundle interface bandwidth. This percentage is then applied to every member link on the egress, or to the local bundle instance on ingress.

• guaranteed bandwidth—Provided in percentage and applied to every member link.

• traffic shaping—Provided in percentage and applied to every member link.

• WRED—Minimum and maximum parameters are converted to the right proportion per member link or bundle instance, and then are applied to the bundle.

• marking—Process of changing the packet QoS level according to a policy.

• tail drop— Packets are dropped when the queue is full.

802.1Q VLAN subinterfaces can be configured on 802.3ad Ethernet link bundles. Keep this information in mind when adding VLANs on an Ethernet link bundle:

• The maximum number of VLANs allowed per bundle is 4000.

• The maximum number of bundled VLANs allowed per router is 128000.

Note	The memory requirement for bundle VLANs is slightly higher than standard physical interfaces.

To create a VLAN subinterface on a bundle, include the VLAN subinterface instance with the interface Bundle-Ether command:

interface Bundle-Ether instance.subinterface

After you create a VLAN on an Ethernet link bundle, all physical VLAN subinterface configuration is supported on that link bundle.

These steps provide a general overview of the link bundle configuration process. Keep in mind that a link must be cleared of all previous network layer configuration before it can be added to a bundle:

1. In global configuration mode, create a link bundle. To create an Ethernet link bundle, enter the interface Bundle-Ether command.

2. Assign an IP address and subnet mask to the virtual interface using the ipv4 address command.

3. Add interfaces to the bundle you created in Step 1 with the bundle id command in the interface configuration submode. You can add up to 32 links to a single bundle.

Note	A link is configured to be a member of a bundle from the interface configuration submode for that link.

The Cisco IOS XR software supports nonstop forwarding during failover between active and standby paired RSP cards. Nonstop forwarding ensures that there is no change in the state of the link bundles when a failover occurs.

For example, if an active RSP fails, the standby RSP becomes operational. The configuration, node state, and checkpoint data of the failed RSP are replicated to the standby RSP. The bundled interfaces will all be present when the standby RSP becomes the active RSP.

Note	Failover is always onto the standby RSP.

Note	You do not need to configure anything to guarantee that the standby interface configurations are maintained.

When one member link in a bundle fails, traffic is redirected to the remaining operational member links and traffic flow remains uninterrupted.

A bundle is a group of one or more ports that are aggregated together and treated as a single link. The different links within a single bundle can have varying speeds, where the fastest link can be a maximum of four times greater than the slowest link. Each bundle has a single MAC, a single IP address, and a single configuration set (such as ACLs or QoS).

The router supports bundling for these types of interfaces:

• Ethernet interfaces

• VLAN subinterfaces