• PUB Master Student: Mihaela Vlad (INRIA Master Internship)
• INRIA Advisor: Alexandra Carpen-Amarie

This work addresses how monitoring can be used in keeping track of user activity inside a storage system. It discusses the useful data that should be collected by the monitoring tools in order to present the user with a detailed and meaningful image of the storage system and of the data inside. A key point is to identify how the data must be processed in order to obtain a real-time image of the integrity of information within the system and of the users creating it.

The project also examines the monitoring challenges raised by a large-scale storage system, such as the large number of nodes, the fine-grained striping of the data over the storage nodes and the heavy concurrent access to data, as it is the case with BlobSeer. The MonALISA system is a monitoring framework able to meet these challenges, and a user accounting system for BlobSeer can rely on it for gathering the required information.

This work is based on some previous research addressing how monitoring can be used in tracking the activity inside BlobSeer: a visualization tool of the WRITEs and READs within the system and a centralized user accounting system based on the data collected in the previous step. Both of them use a centralized storage for the monitoring data, and the information about operations was indexed by BLOBs. What we propose in addition to their work is adapting the information received into a distributed architecture for storing and processing user-related data, and also a module that uses the results obtained to compute a list of what we call malicious clients.

More details: Thesis