1998 - 1999 CS Annual Report                                                        Researchers
choices.gif (4488 bytes)
 

Robbert van Renesse

Senior Research Associate
rvr@cs.cornell.edu
http://www.cs.cornell.edu/home/rvr/

PhD Vrije Univ., Amsterdam 1989

I'm directing, with Ken Birman and Werner Vogels, a new project called Spinglass. The project investigates a new direction in scalable group communication. The idea is to use gossiping between the group members to have them approximate a shared state. Gossiping protocols can be
analyzed using epidemiological theory, resulting in probabilistic guarantees on the accuracy of the shared state. These protocols scale much better than traditional group communication protocols. Most recently I have developed an extremely scalable resource monitoring and location system based on such protocols. 

I'm also involved in the NuPrl/Ensemble effort, a joint effort with Bob Constable's Nuprl group, which intends to harden the core Ensemble group communication sys tem and a large set of Ensemble's micro-protocols. We are specifying the behavior of group communication protocols in I/O automata, and are using Nuprl to construct mechanical proofs of communication properties.
Our most concrete results thus far are optimizations: using partial evaluation we are able to generate optimized common paths from the original Ensemble code that perform within 50% of the best hand-optimized code, and come with proofs of correctness. 

Finally, I'm also working in the Tacoma project, together with Fred Schneider and Dag Johansen. Tacoma is developing secure mobile code technology, but the techniques that we are developing turn out to be applicable in general distributed systems. We are currently developing a customizable file system, in which files can be dynamically wrapped with layers of code. Such
code could, for example, be inserted by an intruder detection system. We are also developing a new proactive secret sharing protocol, and a secure clock synchronization protocol. Closing the loop, both protocols use secure gossiping. 

Professional Activities

  • Associate Editor: IEEE Trans. Distributed and Parallel Systems 

  • Vice-President, R&D: Reliable Network Solutions, Inc. 

  • Program Committees: IEEE Int. Conf. Distributed Computing Systems 1999; Middleware `98

Panels 
  • Adapt to survive. 8th ACM SIGOPS European Workshop, Sintra, Portugal, Sept. 9, 1998. 
  • Policies for new network forms. ARPA HCN Meeting, Rockport, MA, June 12, 1998. 
Publications 
  • PEX: A partitionable execution service for a network of heterogeneous processors.
    Proc. of the 29th International Symposium on Fault-Tolerant Computing (June 1999) (with T. Hickey).  
  • What TACOMA taught us. Mobility: Processes, Computers, and Agents (D. Milojicic, F. Douglis and R. Wheeler, eds.), Addison Wesley Publishing Company and the ACM Press, (Apr. 1999) (with D. Johansen and F. Schneider).  
  • Specifications and proofs for Ensemble layers. 5th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, LNCS 1579, (R. Cleaveland, ed.), Springer Verlag, (March 1999), 119-133 (with J. Hickey and N.Lynch). 
  • Operating system support for mobile agents. Readings on Agents, (Michael N. Hunhns
    and Munindar P. Singh, eds.), Morgan-Kaufmann Publishers Inc. (1988) (with D. Johansen
    and F. Schneider).  
  • Goal-oriented programming, or composition using events, or threads considered harmful.
    Proc. of the Eighth ACM SIGOPS European Workshop (Sept. 1998).  
  • Six misconceptions about reliable distributed computing. Proc. of the Eighth ACM
    SIGOPS European Workshop     (Sept. 1998) (with W. Vogels and K. Birman).  
  • A gossip-based failure detection service. Proc. of Middleware '98 (Sept. 1998) (with Y.
    Minsky and M. Hayden).  
  • Building adaptive systems using ensemble. Software -Practice and Experience (Aug 1998)
    (with K. Birman, M. Hayden, A. Vaysburd, and D. Karr).