Loopback Interfaces: Why Every Network Device Needs One

A loopback interface is a logical interface that exists as long as the device does, regardless of the state of any physical port. Juniper sums up the value in one line: physical interfaces are removed or re-addressed when the topology changes, whereas the loopback address never changes. That property is what makes it usable as the identity of a device for management, routing protocol session establishment and filtering.

What a loopback buys you

• A stable device identity. Ping, SNMP, SSH and syslog references point at an address that does not move when a link is re-cabled.
• Router ID selection. If no router ID is explicitly configured, the loopback address is chosen; otherwise the device falls back to the lowest IP of any interface in an up operational state - which can change silently.
• iBGP and protocol sessions. Internal BGP peering between loopbacks survives the failure of any single path, and the protocol relies on the underlay to provide reachability.
• Management filtering. Because the loopback is a known, unique /32, access lists and firewall filters protecting management systems are trivial to write and audit.
• Troubleshooting. MPLS ping and some diagnostic commands require a loopback address to work correctly.

Configuration on two platforms

! Cisco IOS / IOS XE
interface Loopback0
 description Router ID and management
 ip address 10.255.0.1 255.255.255.255
 ipv6 address 2001:db8:255::1/128

! Juniper Junos
set interfaces lo0 unit 0 family inet address 10.255.0.1/32
set interfaces lo0 unit 0 family inet6 address 2001:db8:255::1/128
set interfaces lo0 unit 0 family inet address 10.255.0.1/32 primary

Junos requires the loopback to be configured with a /32 mask because the Routing Engine is essentially a host, and it maintains a separate internal loopback instance (lo0.16384) so that a filter applied to lo0.0 cannot disrupt internal traffic. If more than one address is configured, marking one as primary determines the default source address used by the Routing Engine - a small setting with large consequences for traceroute output and ACL design.

Advertising the loopback

! OSPF on Cisco
router ospf 1
 router-id 10.255.0.1
 network 10.255.0.1 0.0.0.0 area 0

! OSPF on Junos
set routing-options router-id 10.255.0.1
set protocols ospf area 0.0.0.0 interface lo0.0 passive

Keep the loopback passive in the IGP: there is no neighbour to form across it, and marking it passive prevents hello packets from being generated needlessly while still advertising the prefix.

Address planning for loopbacks

ISP practice converges on a dedicated, contiguous block for loopbacks, allocated either geographically or per point of presence, and never carved out of the same space as link addressing. The block should be large enough that adding a device somewhere in the network does not force a re-plan, and it should be summarised at area and AS boundaries like any other prefix. Recording the loopback-to-hostname mapping in forward and reverse DNS is what makes the loopback useful operationally rather than just architecturally - being able to type a router's name into traceroute depends on it.

Common mistakes

• Using a physical interface address as the BGP or OSPF router ID. The session may survive, but the identity changes when the topology does.
• Filtering the loopback subnet too aggressively and breaking iBGP peering after an IGP change.
• Configuring the loopback with a subnet mask rather than a host address, which can advertise an entire prefix and cause forwarding surprises.
• Forgetting IPv6. The same argument for a stable identity applies to the ::1-style unique local or globally unique /128 address; see our IPv6 address planning guide.

Related reading on this site: BGP session hardening with GTSM, Junos BFD timers and liveness detection and automated configuration backup with Oxidized.

原文链接：https://www.juniper.net/documentation/us/en/software/junos/interfaces-fundamentals/topics/topic-map/loopback-interfaces.html