Understanding LISP (Locator/ID Separation Protocol)

Introduction:

In today’s ever-expanding network landscape, the need for scalability and mobility is paramount. Traditional routing protocols often struggle to meet these demands efficiently. Enter the Locator/ID Separation Protocol (LISP), a network-layer-based protocol designed to optimize scalability while enhancing mobility. LISP is a routing architecture that provides new semantics for IP addressing. It decouples the identity of an endpoint from its location into two separate namespaces; Endpoint Identifiers (EIDs) and Routing Locators (RLOCs). It is this approach that drives the benefits of LISP; routing scalability, mobility, multi-homing, simplified and flexible routing operations,  capacity planning etc. While LISP was initially developed by Cisco, it is important to note that it operates as an open standard rather than a proprietary solution, as defined in RFC 6830. The primary driver for developing LISP was to address scaling challenges of internet routing and addressing, which was a problem statement produced by the October 2006 IAB Routing and Addressing Workshop (RFC 4984). LISP currently has use-cases in campus networks (used in Cisco SD-Access), data centers networks, WAN etc.

In this article, we delve into the characteristics of traditional routing, explore how LISP addresses these challenges. We will discuss LISP architecture components in another article.

Traditional Routing Challenges:

Traditional routing protocols bind/couple endpoint and location information into a single address (remember the host portion, network portion construct in a single IP when learning IP addressing … ). This limits mobility, requiring readdressing when endpoints relocate. Moreover, routers exchange prefixes proactively using a push model, consuming resources and complicating traffic engineering and multi-homing. This approach leads to several challenges:

1. Mobility Limitations: There are many applications that require hosts to change their points of attachment to the network while maintaining uninterrupted communication to other nodes on the network. Hosts maintaining the same identity regardless of where they move on the network is essential for seamless mobility. In traditional routing schemes, because an endpoint’s IP address and location is bound in a single address, the endpoint is effectively tied to the topology (the endpoint IP reflects both the endpoint and its location in the topology), and a move of the endpoint to a different location in the network requires readdressing of the endpoint, thereby limiting mobility. This introduces overhead and complexity, hindering seamless mobility for devices such as mobile phones, laptops, and IoT devices.

2. Prefix Exchange Overhead: Traditional routing protocols, such as Border Gateway Protocol (BGP), exchange routing prefixes between routers regardless of whether a router has hosts in an attached network that need to communicate with hosts in received prefixes. This widespread propagation of prefixes leads to increased overhead, excessive memory consumption, and inefficient routing table management.

3. Device Resource Requirements: Routers participating in traditional routing protocols require significant resources, including Ternary Content Addressable Memory (TCAM), Random Access Memory (RAM), and CPU processing power, to store and process routing information. As networks grow larger and more complex, the resource demands placed on routers escalate, potentially leading to performance degradation and operational inefficiencies. This growth pattern impacts capacity planning and the overall cost of maintaining the network as larger devices are required to accommodate the prefixes and support routing operations.

For example, consider the network topology below; each router has a set of hosts/prefixes beneath it with BGP running between all the four routers to advertise these prefixes into the network. The outcome of this approach is that all four routers receive all prefixes in the network and keep these in their routing table regardless of if they have local hosts that need to reach any of these remote/received prefixes or not.  This is the push model of traditional routing, which results in inefficiencies from routing operation perspective and device resource utilization. This challenge is compounded with the movement of hosts between routers i.e. mobility. Because the IP address in traditional routing reflects both the entity and its location in a single construct, moving a host from R1 to for example R2 will require readdressing of the host impacting seamless mobility. Also, depending on how the host prefix is advertised, this will mean a readvertisement (push) of this new information to all routers in the routing domain with every host move.

How LISP Addresses These Challenges:

LISP introduces innovative solutions to address the shortcomings of traditional routing protocols:

1. Separation of Endpoint Identifiers and Routing Locators: LISP replaces IP addresses with two new constructs/namespaces: Endpoint Identifiers (EIDs) and Routing Locators (RLOCs). EIDs are assigned to hosts, while RLOCs are assigned to routers. By decoupling endpoint identity from its location within the network topology, LISP enables seamless mobility without the need for readdressing. This separation optimizes routing system scalability scalability and simplifies network management.

2. Dynamic Mapping System: Unlike traditional routing protocols, where routers exchange prefix information proactively, LISP routers query a dynamic mapping system only when traffic needs to be sent to a specific EID. This on-demand mapping retrieval reduces overhead and minimizes the dissemination of unnecessary routing information throughout the network. LISP’s dynamic mapping system improves routing efficiency and resource utilization while accommodating mobility and scalability requirements.

Following our referenced network topology again, let us assume R4 is now the mapping system (we will discuss more in the article that looks at the LISP architecture). With the decoupling of the host IP address and its location into separate entities i.e  Endpoint Identifiers (EIDs) and Routing Locators (RLOCs) respectively, every router does not need to have knowledge of the EIDs and their location, only the mapping system needs to keep this EID-to-RLOC mappings. When routers need to send traffic to a remote prefix, the mapping system will be queried on-demand for the location of that EID. When a host on R1 moves to R2, the EID of the host is maintained, and the mapping system updated with the new location.

Conclusion:

In an era where network scalability and mobility are critical, traditional routing protocols fall short. LISP offers a paradigm shift by separating location and identity, optimizing routing system scalability, and enhancing mobility. In the next article, we will discuss the architecture and components of LISP.

References:

– [RFC 6830](https://datatracker.ietf.org/doc/html/rfc6830): “The Locator/ID Separation Protocol (LISP)”

– [RFC 4984](https://datatracker.ietf.org/doc/html/rfc4984): “Report from the IAB Workshop on Routing and Addressing”