The nano datacenter (NADA) architecture of the NANODATACENTERS project is a new distributed computing paradigm that relies on “nano” sized interconnected data centres placed around the edges of the network. The architecture aims to address the concerns and shortcomings of the currently deployed monolithic datacenters.

Architectures for content distribution tend to fall into one of two diametrically opposed categories: either totally centralised monolithic datacenters, or else highly distributed peer-to-peer (P2P) nodes operating independently. Both of these paradigms have their particular advantages and disadvantages. For instance, monolithic datacenters are associated with high energy consumption costs, scalability problems and maintenance issues. However, such a centralised architecture does provide enhanced security and can adhere more easily to Service Level Agreements (SLAs) due to its well-controllable nature. On the other hand, a P2P architecture handles better the energy, scalability and maintenance issues faced by the centralised architecture. However, the P2P architecture faces problems with security, selfishness in the system due to the presence of selfserving nodes, and a lack of service guarantees, due to the autonomous nature of the elements in the architecture.

The NADA system consists of a highly distributed arrangement of devices at the edges of the network that are capable of hosting and delivering content. These devices are owned by a single authority (for instance an ISP) and rely on a P2P communication infrastructure. The combination of highly distributed resources together with the fact that all of the resources are under the
responsibility and control of one single authority, helps address the concerns raised earlier.

This document describes the NADA architecture. It begins by presenting the issues which the NADA architecture addresses. Each of these issues is then explained by presenting a high-level overview of the system components along with the underlying reasons why certain design decisions were taken. Each component is then described in detail. Finally, a case-study is presented of how an application such as a Video-on-Demand service can use NADA.