Real Time and Forensic Network Data Analysis Using Animated Combined Visualizations Sven Krasser Gregory Conti Julian Grizzard Jeff Gribschaw Henry Owen Georgia Institute of Technology Overview of Visualization packet size 255.255.255.255 65535 color: protocol time now age destination port source IP address ol oc e ot ag p r ss: r: lo ne co i g h t br 0.0.0.0 color: protocol 0 now time packet size age Overview of Visualization packet size 255.255.255.255 65535 color: protocol time now age destination port source IP address ol oc e ot ag p r ss: r: lo ne co i g h t br 0.0.0.0 color: protocol 0 now time packet size age Motivation • High level analysis - low level discovery • Complement Ethereal by providing big picture context • TIVO for Network Traffic • Dealing with customers • Network behavior / Intruder behavior • Support Honeynet log analysis • Not real-time intrusion detection (yet) System Design • real time packet capture and forensic playback • navigate forwards and backwards in dataset • 3D and 2D views • Open GL and commodity hardware (P4 2.5GB) • Parallel coordinate plot adjacent to two animated displays Overview and Detail Routine Honeynet Traffic (baseline) Slammer Worm Constant Bitrate UDP Traffic Port Sweep Attempted HTTP Attack… Attempted HTTP Attack… (zoom) Compromised Honeypot Attacker Transfers Three Files… campus network Inbound Campus Traffic (5 seconds) Campus Network Traffic (10 msec capture) inbound outbound botnet visualization Combined botnet/honeynet traffic System Performance System Performance Conclusions • • • • • Combining of visualization techniques Open GL and commodity hardware Significant analyst performance gains Interaction techniques Distinct visual signatures – Smart Books • Tipping point on high volume networks – Honeynet /CTF analysis possible now – Prefiltering required for general purpose use Future Work • Semantic zoom – packets -> flows -> application/protocol specific • Work through slices of network traffic – allow user to focus on what is interesting • Maximize customization and interaction – Filtering and encoding – All fields • Multiple data streams • Knowledge discovery • Help highlight what is interesting • Easily drop in different windows on network traffic – look at traffic from different perspectives • Evaluation Demo of tools Acknowledgements • Charles Robert Simpson for providing NETI@home packet capture source code • David Dagon for for providing the botnet data Questions? Sven Krasser sven@ece.gatech.edu Gregory Conti conti@cc.gatech.edu Julian Grizzard grizzard@ece.gatech.edu Jeff Gribschaw jgribsch@ece.gatech.edu Paper Henry Owen henry.owen@ece.gatech.edu Image: http://altura.speedera.net/ccimg.catalogcity.com/210000/211700/211780/Products/6203927.jpg