Unnumbered IPv4 Interfaces

blog » series » Unnumbered IPv4 Interfaces

IPv4 was designed to be used on large hosts connected to a third-party proprietary WAN network and thus assumed that an IP subnet would be assigned to every (edge) network segment. However, when we started deploying IPv4-only networks, we quickly discovered scenarios where this assumption wasted address space or made deploying the desired solution impossible.

Vendors tried to circumvent that limitation with various unnumbered IPv4 interface solutions, starting with point-to-point links and dialup scenarios and eventually ending with unnumbered Ethernet interfaces.

We don’t have a similar problem in the IPv6 world as every IPv6-enabled interface always gets a link-local address.

These blog posts will help you understand why IPv4 uses interface addresses, what we can gain with unnumbered IPv4 addresses, how they work over Ethernet links, and how they interact with routing protocols.

Building Unnumbered Ethernet Lab with netlab
Want to try out unnumbered Ethernet interfaces on a dozen different platforms? This is how you can easily build your lab with netlab.
Back to Basics: Do We Need Interface Addresses?
In the world of ubiquitous Ethernet and IP, it’s common to think that one needs addresses in packet headers in every layer of the protocol stack. However, that’s just one option and not exactly the best one in many scenarios.
Back to Basics: The History of IP Interface Addresses
This blog post describes some of the history behind the decision to use interface (not node) addresses in IPv4.
Back to Basics: Unnumbered IPv4 Interfaces
The decision to use interface addresses in IPv4 caused interesting problems in dial-up scenarios, resulting in numerous implementations of unnumbered IPv4 interfaces.
Packet Forwarding and Routing over Unnumbered Interfaces
IP routes usually require a next-hop IPv4 address. How does that work in designs where the interfaces do not have IPv4 addresses or where the interface addresses of adjacent routers are not in the same subnet?
Unnumbered Ethernet Interfaces
Unnumbered IPv4 interfaces were traditionally available for point-to-point (and dial-up) links. Can we make them work with Ethernet- or VLAN interfaces?
Unnumbered Ethernet Interfaces, DHCP Edition
We can make unnumbered Ethernet interfaces work with creative use of static routes pointing to interfaces (not next hops), but how does that work in designs where the clients attached to an unnumbered Ethernet interface use DHCP?
Running OSPF over Unnumbered Ethernet Interfaces
Can we run OSPFv2 over unnumbered point-to-point links? Yes, that’s been defined in the very early OSPFv2 RFCs. But what does it take to make OSPFv2 work over unnumbered Ethernet interfaces?
Running IS-IS over Unnumbered Ethernet Interfaces
OSPFv2 can run over unnumbered point-to-point links. Some IS-IS implementations are better than that and can run over unnumbered multi-access segments (for example, Carrier Ethernet E-LAN service).
BGP Unnumbered Duct Tape
We could always run BGP across unnumbered links if we used an IGP or static routes to propagate the loopback addresses of adjacent nodes. Recently, many vendors started supporting another solution: run EBGP over IPv6 LLA and exchange IPv4 prefixes over that EBGP session using the next-hop encoding specified in RFC 5549. How exactly does that work?
OSPF and ARP on Unnumbered IPv4 Interfaces
Time for a deep dive: how do OSPFv2 and ARP interact to make routing work over unnumbered IPv4 interfaces?
OSPF ECMP with Unnumbered IPv4 Interfaces
Can we do equal-cost multipathing over a set of unnumbered point-to-point links? Of course, that worked since the early days of OSPFv2.
Review: Unnumbered Interfaces in netlab
Back to netlab: it supports IS-IS and OSPFv2 on unnumbered interfaces on almost a dozen platforms. It’s time to kick the tires and figure out how unnumbered IPv4 interfaces work on your favorite platform.
Sidebar