Skip to content

Latest commit

 

History

History
58 lines (50 loc) · 3.02 KB

File metadata and controls

58 lines (50 loc) · 3.02 KB

Chapter |Fed|: Federating Networks

Having looked at the operation of switches and the routing algorithms that allow us to determine best paths through large networks, we are well on our way to being able to build a network of global scale. However, there is another aspect of the Internet that has enabled it to scale to connect the majority of the world's population: the ability to interconnect independently built and operated networks.

This results in a global internetwork that has no central point of control and yet manages to deliver a reasonably consistent service to all those devices and the applications running on them. Now that the Internet has been around for more than 50 years it is easy to lose sight of the design decisions and challenges that had to be tackled to achieve this remarkable feat. In this chapter we examine the problem of federating networks: taking networks that are run by independent operators using widely varying underlying technologies and connecting them together to provide a useful, global packet delivery service.

Federating autonomous networks helps scale the approach to operations, since each organization that manages a network can make its own independent decisions about how to manage that network. It also gives us a handle on how to scale routing to calculate paths across a global network by breaking routing into two parts: routing among networks and routing within individual networks. At the same time we need to find a way to address billions of devices, a problem that turned out to be bigger than first envisioned by the Internet's designers.

Compounding the issue of scale is the problem of heterogeneity. The original Internet connected the ARPANET, a pioneering packet-switched network, with some other early networks that used radio and satellite technology. One of the early design decisions was that the other networks would not be modified to "fit" into the Internet; the Internet would have to be able to accommodate all these different technologies. That has proven to be one of the pivotal decisions that has allowed the Internet to succeed as it has accommodated new technologies ranging from multi-gigabit optical fiber to 5G wireless—technologies that were not even on the horizon at the time of the Internet's design.

This chapter addresses the challenges of heterogeneity and scale, which are substantial, but it is important to not lose sight of the bigger picture: that supporting autonomy is the foundational premise. Fully embracing autonomy means that the organizations responsible for the underlying networks retain the right to set policies about how their networks are used. The Internet's approach to specifying and enforcing such policies is a complex topic in its own right, and one we address in the next chapter.