Clean-Slate Information-Centric Publish Subscribe Networks

Clean-Slate Information-Centric Publish Subscribe Networks

Laura Carrea (University of Essex, UK) and Raul Almeida (University of Essex, UK)
Copyright: © 2012 |Pages: 28
DOI: 10.4018/978-1-4666-1613-4.ch013


The Internet architecture of today does not seem suited to the current Internet usage, as the application layer is more and more content-centric, while the network layer is ossified around the IP concept. In this chapter, the authors explore a redefinition of the whole Internet architecture where nothing is taken for granted, especially IP addresses. The review focuses on the forwarding and topology components of the EU FP7 PSIRP architecture and on a few of the problematic issues and the ongoing discussions around a pioneering clean-slate design of the way to organize networks.
Chapter Preview


The Internet architecture of today is a packet-based internetworking architecture that was created as an efficient multiplexed utilization of heterogeneous interconnected networks. The introduction of the Internet Protocol suite (TCP/IP) opened up the growth of the Internet and slowly the backbone was privatized and became distributed. Because a central coordinating element is missing, it has been difficult to apply major architectural changes to the Internet and the Internet architecture (i.e. hierarchical routing, TCP/IP, Domain Name System (DNS)) has remained the same since the 1980s when it was created (Jacobson, 2006a). Since then, only incremental improvements have been introduced to supply new services reducing the management costs of the network: the Classless Inter-Domain Routing (CIDR) (Fuller et al., 1993) was proposed to slow the growth of routing tables on routers across the Internet, and to slow the rapid exhaustion of IPv4 addresses; the Border Gateway Protocol (BGP) (Rekhter, 1995) was introduced to mirror the business relationship between providers and later extended for large scale deployments; the Multi-Protocol Label Switching (MPLS) (Rosen et al., 2001) was introduced to improve the performances of IP routers. Later, Virtual Private Networks (VPN) and, recently Carrier Ethernet have appeared as new data services. Moreover, other solutions have been deployed for issues of Internet design in an open commercial environment such as Network Address Translation (NAT) boxes, which offer limited protection for unwanted traffic fracturing network connectivity and which have extended address spaces (Touch, 2002), (Handley, 2006) and Mobile IP (Perkins, 2002), which offers mobility to the host, using network indirection points.

However, despite all those efforts, Internet is considered as ossified (for example (Handley, 2006), (Anderson et al., 2005)), as all the solutions are considered as patching approaches based on ad-hoc extensions and overlays.

The original design of the Internet is centered on best-effort delivery between network-attached devices, forming the base of the concept of the Internet Protocol (IP) address. Every host has a unique IP address which acts as a location (where) utilized for routing purposes and at the same time as an identifier (who) of the host. The fact that the IP address has these two different functions is considered the root of many of the limitations of today’s Internet architecture (Meyer et al., 2007) and a split is considered necessary.

In the past and still today, efforts for the future Internet have been based on revisiting single concepts, such as the IP locator/identifier, to improve end-host reachability, end-to-end security, mobility and routing issues without questioning the host centricity of the communication. Still, the Internet moves a datagram in a best effort manner independently of the semantics and the purpose of the data transport.

Complete Chapter List

Search this Book: