Shrideep Pallickara, Jaliya Ekanayake, Geoffrey Fox Community Grids Lab Indiana University Collaborative Analysis of Distributed Data Applied to Particle Physics Experiments
NaradaBrokering: Q UICK S UMMARY Content distribution infrastructure for data streams Framework for development of distributed systems Funding Sources Two grants from the National Science Foundation Two grants from the United Kingdom’s OMII Recent STTR grant from the Department of Energy Code base specifics: Open Source, Version ,000 lines, 1425 classes, 157 packages
Stream dissemination: H IGHLIGHTS Fine-grained selectivity within a stream Regular Expressions, SQL, XQuery & XPath queries Stream jitter reduction Time-ordering of streams Support for multiple transports TCP, UDP, Multicast, SSL, HTTP, Parallel-TCP
Information Assurance: S ECURITY Restrict discovery of streams Control who (and for how long) can generate & consume streams Encrypt streams to prevent eavesdropping Digitally sign streams to detect tampering Cope with Denial of Service attacks
Information Assurance: F AULT T OLERANCE Cope with node failures Guaranteed delivery despite failures Support for recovery from failures Fine tune redundancy Scalable tracking of resources within system
Clarens Secure, high-performance portal Ubiquitous access to data & computational resources Uses ROOT for analysis of particle physics data Generated by the Compact Muon Solenoid (CMS) detector at CERN
NB Clarens: S ALIENT F EATURES Stand-alone application converted into a collaborative one Participation predicated on authorization No limits on number of participants Data-driven distribution of computations
Demo of Collaborative Analysis
In the works Dynamic real-time distribution and balancing of computation loads. Incorporate monitoring services Secure collaborative sessions Streams within session will be encrypted Streams will be digitally signed for tamper evidence