NetFlow data is routinely captured at the border of many enterprise networks. Although not as rich as full packetcapture data, NetFlow provides a compact record of the interactions between host pairs on either side of the monitored border. Analysis of this data presents a challenge to the security analyst due to its volume. We report preliminary results on the development of a suite of visualization tools that are intended to complement command line tools, such as those from the SiLK Tools, that are currently used by analysts to perform forensic analysis of NetFlow data. The current version of the tool set draws on three visual paradigms: activity diagrams that display various aspects of multiple individual host behaviors as color 1 coded time series, connection bundles that show the interactions among hosts and groups of hosts, and the NetBytes viewer that allows detailed examination of the port and volume behaviors of an individual host over a period of time. The system supports drill down for additional detail and pivoting that allows the analyst to examine the relationships among the displays. SiLK data is preprocessed into a relational database to drive the display modes, and the tools can interact with the SiLK system to extract additional data as necessary.

No abstract

Today's distributed computing environments, like Energy Control Systems, lack a common and adaptive notion of trust and are vulnerable to a wide range of attacks from complex threats. These threats on our control systems are distributed, decentralized, dynamic, and operate over multiple timescales. Threats may also result from structural weaknesses in system designs that permit exploitation by insiders working inside globally trusted service providers. Although approaches such as Trusted Computing are part of the solution, we argue that a layered notion of distributed trust is required to effectively address the end-to-end security needs of these systems.