Distributed Storage Strategies for IP Traffic Traces

State: completed by Thierry Kramis

The key aim of this thesis is the design, prototypical implementation and performance evaluation of different distributed storage strategies for IP netflow traces. A basic design for a P2P-based storage platform for IP flow records already exists and basic functionality is implemented. This work shall concentrate on developing different strategies for storing IP flow records on top of this platform. The IP flow records are stored on the platform based on their flow IDs. A flow ID is the result obtained by aplying a hash function to the IP header fields that define the flow. A storage strategy for IP flow records consists of designing hash functions and storage rules that allow to group together flows with similar characteristics. Different storage strategies highly impact the message overhead and processing requirements of storing and retrieving flow records. During the thesis the student shall design and implement different strategies for distributed storage of IP flow records and compare their efficiency and performance.

Final Report

40% Design, 60% Implementation
Java programming, P2P knowledge

Supervisors: Prof. Dr. Burkhard Stiller

back to the main page