Advances in efficient packet forwarding techniques have been central to continuously moving traffic smoothly through the Internet at increasing rates. Much work has been invested in data structures and algorithms for packet forwarding and classification. Research in the design of forwarding table compacting techniques has been a continuum since the early 90’s,
and still goes on, yielding novel compact representations for structured graphs such as tries, new algorithms and data structures for IP lookups, packet classification, and advances in high-speed memory technologies among others.

Content-oriented network architectures are characterized by introducing new namespaces for content objects. A common property of the proposed naming schemes is relying on flat identifiers (e.g., 256-bit hash values) and/or long, non-fixed size URL-like names (e.g., TRIAD, CCN) to uniquely identify single pieces of content. Other network architectures that separate identifiers from locators or aiming at scalable Ethernet designs, face similar challenges when handling packets carrying flat identifiers. A flat naming scheme simplifies address administration or content identification but is hard to scale due to the lack of aggregation capabilities. Structured identifiers (e.g., NDN) are also hard to handle at wire speed due to the challenges of performing LPM-like lookup operations on arbitrary long identifiers resulting from non-fixed size components.

Similar to the advances in algorithms and data structures that enabled the feasibility of high-performance IP routers, we surmise that new enablers in the forwarding plane may be fundamental to the realization of content-oriented networks. More specifically, we expect probabilistic techniques to play a key role to guide the construction of data structures well-suited for the requirements of packet forwarding in content-oriented networks. 

Motivated by the needs of content-oriented networking proposals, we have explored new approaches to the fundamental trade-offs of packet routing to provide forwarding services with scalability, multicast-friendliness and security in mind. The main idea behind compact forwarding is taking a probabilistic approach to the problem of packet forwarding in networks centered on content identifiers rather than traditional host addresses. 
Due to the lack of aggregation capabilities of flat labels and the compact forwarding goal of seeking the minimal information base to deliver packets at scale, we have dived into solutions based on error-prone probabilistic data structures providing lossy compression functionality. 

A fundamental question explored is where to place the packet forwarding state, in network nodes or in packet headers? Solutions for both extremes are proposed. In the SPSwitch, approximate forwarding state is kept in network nodes. In LIPSIN, the state is carried in the packets themselves. Both approaches are based on probabilistic packet forwarding functions inspired by variations of the Bloom filter data structure. The approximate forwarding state comes at the cost of additional considerations due to the effects of one-sided error-prone data structures. By exchanging correctness (traduced in forwarding efficiency penalties) for space/memory time requirements (traduced in reduced forwarding information base in packet headers and network nodes), new spots in the design space can be explored.