Towards Network Awareness

Network and system administrators need to analyse network traffic for maintenance, security, and planning purposes. The volume of data on modern networks, however, make such analysis extremely difficult using existing open source tools. In this paper we argue that administrators need tools that will allow them to be more aware of the state of their networks, and we describe our vision for tools that would support such "network awareness" by analysing and visualising packet aggregations that are defined by both packet headers and payloads.

As a first step towards such tools, we have developed a library called qcap, a framework for packet and stream reconstruction that allows applications to tap packets at all layers of the network stack: from network, to transport, to the application layer. qcap is fast, able to process network data at speeds of 120 megabytes per second on commodity hardware; it is easy to use, providing a simple API that requires only a few lines of code to perform complex parsing tasks; and it is extensible, using BNF-like grammars to describe TCP protocols. We believe that qcap can provide the foundation for tools that will support greater network awareness for system administrators.