New IP: Enabling the Next Wave of Networking Innovation

Introduction

Current Internet technologies, with the Internet Protocol (IP) as its bedrock, have been a tremendous success. Originally devised as a way to interconnect computers over long distances in a highly resilient, decentralized manner that would be able to withstand outages of links and nodes, IP has evolved in a breathtaking fashion ever since. Today, the Internet interconnects not just a handful of computers but billions of devices (including “Things”) with an exorbitant volume of traffic growing at a breakneck pace. In addition, the Internet is able to support services that did not exist at the time of its conception – such as the World Wide Web or Social Media – and that would have been deemed impossible to support at the time – such as “real time” services like voice and interactive video. Throughout its evolution, the Internet has also been a driving force enabling network convergence, allowing previously separate custom networks to be integrated into a single infrastructure providing significant economies of scale. This made networks much more economical to operate, further contributing to their increasing pervasiveness.

It is thus fair to say that the Internet has come a long way, morphing from a niche novelty into a societal backbone that supports an ever-expanding set of services. Internet technology centers on IP (Internet Protocol) and the TCP/IP protocol suite in the data plane, as well as related mechanisms such as Quality of Service (QoS). While Internet technology has seen many extensions, at the core of its success are still its original principles based on best-effort, hop-by-hop forwarding. These principles do not provide any hard guarantees about whether packets will arrive within a certain time, or even in their original order, or even arrive at all, but they offer simplicity in terms of their implementation and resilience to perturbations experienced in the network. This does not mean that there have not been plenty of advances. Switching technologies like MPLS provide enforcement of homogenous traffic forwarding policies for each and every packet of a given flow. Traffic Engineering (TE) techniques like MPLS-TE with Path Computation Element (PCE) provide some guarantees about the compliance of a path with traffic requirements and fast-reroute capabilities at the network layer. However, none of these technologies is able to provide actual end-to-end latency gurantees or in general more Service Level Agreement (SLA) aware connectivity. In general it is left up to other layers to deal with any fallout of, for example, lost packets.

However, the question is whether the same principles will be applied forever, able to support any and all technical requirements of an ever-growing array of services and converging even more networks until all networking infrastructure can be consolidated, or whether foundational barriers will eventually be encountered that will prove hard if not impossible to overcome. As laid out later in this chapter, in recent years, a number of new requirements and networking use cases have begun to appear which cannot be readily supported by today’s Internet technologies. This is because they exhibit properties that cannot be remedied by simply layering new capabilities on top, even assuming that the issue of Internet ossification can be overcome. Note that Internet ossification refers to the increasing inability to add new services, capabilities, and functions to the existing Internet, due to a variety of factors from adversarial impact of middleboxes to standardization hurdles. The fact that many of these use cases open up the possibility for a whole new wave of innovation for networked applications and connected industries adds to the pressure for answers to the question of how the required capabilities might be supported in the future.