Felix Rauch

Dr. sc. techn. 2003, Dipl. Inf.-Ing. ETH 1997

Note on my new name: After my marriage on February 26, 2004, my official name is Felix Rauch Valenti.
Please read the disclaimer at the end of this page before clicking on any links or using any email-adresses on these pages.

Since may 2004 I am no longer working at the laboratory for computer systems. My new homepage with updated information can be found at the University of New South Wales, Sydney, Australia, where I am now working in the ERTOS program of National ICT Australia (NICTA).

This homepage is therefore no longer maintained.

Shortcuts: Work & projects | publications (dissertation) | student projects | personal activities and other material | disclaimer.

Work and projects

My research interest are systems for data replication, storage, distribution and gathering of large amounts of data, as well as parallel and distributed filesystems for high-speed networks in clusters of PCs. My main focus is on the performance and the architecture of such systems.

Until May 2004 I was a lecturer and postdoctoral researcher at the Laboratory for Computer Systems of the Swiss Federal Institute of Technology (ETHZ). Before, I was a member of the Parallel and Distributed Systems Research Group of Prof. Thomas M. Stricker.

The main umbrella project is named CoPs, which stands for Cluster of PCs. The CoPs-Project investigates architectural and operating system support for parallel and distributed computing on PC clusters interconnected with a Gigabit/sec network. I contributed to the following subprojects of CoPs:

Distributed file systems vs. distributed devices:
This project studies the problem of data gathering in a cluster of PCs interconnected by a high-speed network. The problem being analyzed is how a machine in a cluster, like e.g. a data-base server used for data mining, can access large amounts of data efficiently over a high-speed interconnection network. This type of data transfer is somewhat the opposite of the distribution of larga data sets mentioned below. Two approaches are currently being modelled, analyzed and evaluated: An underlying full-featured distributed file system vs. simple distributed devices that can be accessed with much less overhead. Results will be published at the sixteenth Australasian Database Conference 2005.
Xibalba PC cluster infrastructure:
Our research group was involved in architecting and evaluating the Xibalba cluster, a commodity-PC cluster consisting of 128 1-GHz nodes and used by several research groups in our departement since summer 2001. The cluster runs multiple operating systems which makes it a viable platform for the different needs of the computer-science research community. Our analysis and evaluation of the cost/performance tradeoffs in the network interconnects for the Xibalba cluster are described in a paper presented at the workshop on ommunication architecture for clusters in 2003 and a corresponding technical report. Part of my work was also to install the operating systems on the nodes, to setup a remote booting system, a small maintenance system that runs completely in a RAM disk and to create a system to maintain the installations by cloning them. The experience from the Patagonia project (see below) allowed to re-use the tools Dolly and CloneSys even for this large cluster, thus proving their viability in practice even for larger clusters.
Partition repositories:
The experience with the setup, maintenance and management of the Patagonia cluster and all its different operarting system installations led to the development of partition repositories, which provide a clean abstraction from the operating system configuration state and are an efficient, incremental storage scheme to maintain all necessary partition images for large, versatile clusters of PCs. A research paper presented at the international conference on cluster computing 2000 describes and evaluates partition repositories.
Model and optimization for distributing large data sets in PC clusters:
From the experience with the data-casting tool Dolly derived a model and an optimization of data distribution for large data sets in PC clusters. The model helps to understand and improve the performance of a high-speed multicast operation. It is described and evaluated in a paper presented as distinguished paper at the EuroPar conference in 2000 and in a journal article published in Concurrency and Computation: Practice and Experience.
IT-Talks announcing system:
IT-Talks is a system to submit centrally talks in the range of computer science and associated topics and to automatically publish them on different media. It consists of a back end to store and manage the talks and several front ends to publish the information. The most popular front ends are the LCD touch screens mounted in the IFW building that allow to conveniently browse through upcoming talks and other related information of interest. The system is in use since 2002 with increasing popularity.
Patagonia, a multi-boot and multi-use cluster for research and education:
In this project we were examining the usefulness of a dual-use cluster of PCs for computation and education. We also attacked some common problems for installation, convenient booting, daily usage and maintenance of such a cluster.
Results of the Patagonia project include:
- Dolly, an open-source tool to clone hard disks or partitions over a fast switched network by using a TCP chain between the participating nodes. With Dolly, we were able to clone partitions to any number of nodes with full wire speed on a switched Fast Ethernet.
- CloneSys, a set of open-source scripts that can be used to install and maintain a multi-boot and multi-purpose cluster of PCs. The tools do not depend on an underlaying filesystem on the nodes and can therefore be used to install and update any operating system.
Checkpointing, recovering and migrating processes on different UNIX operating systems:
During my internship at Integrated Systems Engineering AG (Zürich) in summer 1996 I significantly extended and ported a checkpointing library to various UNIX operating systems. The library enabled running processes to migrate between different machines on Sun Solaris, Linux and AIX systems and it allowed to checkpoint and recover running processes on HP-UX and OSF1/alpha machines. The internship report has some more details.

Publications

Dissertation

Felix Rauch: Distribution and Storage of Data on Local and Remote Disks in Multi-Use Clusters of PCs. Diss. ETH No. 15281, Swiss Federal Institute of Technology, December 2003. Published by Hartung-Gorre Verlag Konstanz, Germany, ISBN 3-89649-893-2, Series of Computer Science, ISSN 1611-0943.
There is some more information about the dissertation.

I successfully defended my thesis on september 26 2003 and printed it at the end of december 2003. The progress while writing the thesis is shown on the following pages: page wise, byte wise.

Refereed international journals

Felix Rauch, Christian Kurmann, Thomas M. Stricker: Optimizing the distribution of large data sets in theory and practice. Concurrency and Computation: Practice and Experience, volume 14, issue 3, pages 165-181, april 2002. © John Wiley & Sons, Ltd.
Available from the publisher: abstract, fulltext in pdf.
Available from the authors: abstract, fulltext in pdf.
Ch. Kurmann, F. Rauch and T. Stricker: Speculative Defragmentation — Leading Gigabit Ethernet to True Zero-Copy Communication. Cluster Computing: The Journal of Networks, Software Tools and Applications . Volume 4, Issue 1, page 7-18, March 2001, © Kluwer Academic Publishers
Available formats: abstract, fulltext in pdf, ps.gz.

Refereed publications in international conferences

Felix Rauch and Thomas M. Stricker: OS Support for a Commodity Database on PC Clusters — Distributed Devices vs. Distributed File Systems. Proceedings of Sixteenth Australasian Database Conference (ADC), January 31st - February 3rd 2005, Newcastle, Australia
Available formats: abstract, fulltext in pdf.
Slides: pdf.
Christian Kurmann, Felix Rauch, Thomas M. Stricker: Cost/Performance Tradeoffs in Network Interconnects for Clusters of Commodity PCs Proceedings of Workshop on Communication Architecture for Clusters, April 22, 2003, Nice, France. In conjunction with International Parallel and Distributed Processing Symposium (IPDPS '03).
Available formats: abstract, fulltext in pdf, ps, ps.gz.
Slides: pdf, ps, ps.gz.
F. Rauch, Ch. Kurmann, T. Stricker: Partition Repositories for Partition Cloning — OS Independent Software Maintenance in Large Clusters of PCs. Proceedings of the IEEE International Conference on Cluster Computing 2000, Chemnitz, Germany, Nov 28 - Dec 2, 2000.
Available formats: abstract, fulltext in pdf, ps, ps.gz.
Slides: pdf (94 KB), ps (5.2 MB), ps.gz (1.9 MB).
F. Rauch, Ch. Kurmann, T. Stricker: Partition Cast — Modelling and Optimizing the Distribution of Large Data Sets in PC Clusters. Euro-Par 2000 — Parallel Processing, Arndt Bode and Thomas Ludwig (Editors), Springer, Lecture Notes in Computer Science 1900, ISBN 3-540-67956-1. Presented as distinguished paper at: European Conference on Parallel Computing, Euro-Par 2000, Munich, Germany, August 29 - September 1, 2000.
A similar paper appeared as technical report No. 343 (see below).
Slides: pdf, ps, ps.gz.
Ch. Kurmann, M. Muller, F. Rauch and T. Stricker: Speculative Defragmentation — A Technique to Improve the Communication Software Efficiency for Gigabit Ethernet. Proceedings of the Ninth International Symposium on High Performance Distributed Computing HPDC, Pittsburgh, Pennsylvania, August 1-4, 2000. ISBN 0-7695-0783-2. Pages 131-138.
Available formats: abstract, fulltext in pdf, ps.gz.
Slides: pdf, ps.gz, html.
F. Rauch, C. Kurmann, B. M. Müller-Lagunez, T. M. Stricker Patagonia — A Dual Use Cluster of PCs for Computation and Education Proc. of the second workshop on Cluster-Computing, 25./26. March 1999, Karlsruhe, Germany.
Available formats: abstract, fulltext in pdf, ps, ps.gz.
Slides: pdf, ps, ps.gz.

Correspondence in international journals

Felix Rauch and Thomas M. Stricker: Comments on "Transparent User-Level Process Checkpoint and Restore for Migration" by Bozyigit and Wasiq. Operating Systems Review, ACM, 36(3):8-9, July 2002.
Available formats: abstract, fulltext in pdf.

Technical reports

Christian Kurmann, Felix Rauch, Thomas M. Stricker: Cost/Performance Tradeoffs in Network Interconnects for Clusters of Commodity PCs. Technical report 391, ETH Zürich, Laboratory for Computer Systems, January 2003.
Available formats: abstract, fulltext in pdf, ps, ps.gz.
F. Rauch, Ch. Kurmann, T. Stricker: Partition Cast — Modelling and Optimizing the Distribution of Large Data Sets in PC Clusters. Technical report 343, ETH Zürich, Laboratory of Computer Systems, July 2000.
Available files: abstract, fulltext in pdf, ps, ps.gz.
C. Kurmann, M. Müller, F. Rauch, T. Stricker. Improving the Network Interfaces for Gigabit Ethernet in Clusters of PCs by Protocol Speculation. Technical report 339, ETH Zürich, Laboratory of Computer Systems, April 2000.
Available formats: abstract, fulltext in pdf, ps.gz.

Master thesis, other publications and presentations

Felix Rauch: Zuverlässiges Multicast Transportprotokoll (Reliable Multicast protocol). Diplomarbeit (Masters thesis, in german), march 1997.
Available material: abstract (german), abstract (english), fulltext in pdf (300 KByte, german).
Felix Rauch: Xibalba — A Multi-Boot Cluster for Multiple Uses in Research and Development. Presentation at Linux Days Luxembourg, 2nd October 2002.
Available material: abstract, slides: pdf (500 KByte)
Felix Rauch, Christian Kurmann and Thomas M. Stricker: Using System Commander in Patagonia — A Multi Use Cluster of PCs for Computation, Collaboration and Education. Whitepaper, 2000.
Available material: HTML, pdf.

More publications of our research group can be found on our CoPs publications page.

Student projects

Supervised by me

Diploma theses

Christian Widmer (WS2001/2002): Transparente Zugriffsmethode auf verteilte Festplatten
Christian Wittenhorst (SS2001): Verteiltes Filesystem für einen Cluster of PCs mit Gigabit Ethernet (GigaFS)
Andreas Fleuti (SS2000): Distributed Filesystem for a Cluster of PCs with Myrinet
Eric Dondelinger (WS1999/2000): Elektronischer Veranstaltungskalender für das Departement Informatik
Michael Psarros (WS1998/1999): Verteiltes Filesystem für Hochgeschwindigkeits-Netzwerke
Ivo Sele (WS1998/1999): High-Speed NFS

Term projects

Martin Grüter (SS2001), Partition Cloning
Patrik Stähli (WS2000/2001): Extension of a Performance Counting Library for Pentium Series
Roland Brand (SS1999): Pentium II Performance Counting Library
Andreas Fleuti (SS1999): Distributed Filesystem for a High Speed Network
Rolf Spuler (SS1998): Cluster Administration Tool

Student projects in our group

There is also a complete archive of student projects completed in our group and of course a list of currently open student projects.

Personal activities and other material

My other, personal homepage
From its foundation in 1999 to 2003 I was president of the Swiss Internet User Group, SIUG. Since may 2003 I am vice president.
Since 2000 I am a member of the organizing committee of the annual Swiss Big Brother Awards.
I support the SETI@home project with spare CPU cycles.
My PGP public key (60 KB).
I'm an author for the Symlink newsportal (Slashdot-like, but in german language).
During my work with the cluster in our lab I have collected some informations about Linux.
Some interesting MP3 files.
A few private links.

Warnings / Disclaimers

Because of past events, I must kindly ask you to respect the following disclaimer:

By clicking on some of the links on this page, and then clicking on links on the pages you reach from there, and finaly clicking on some more links, you might eventually reach pages which

have contents which are illegal in some countries
contain copyrighted material
are, depending on your taste, ugly and disgusting to look at
you might not like
have contents I do not agree with

I am sorry, but I can not take responsibility for the whole internet, so use the Internet at your own risk!

Swiss privacy law / Schweizer Datenschutzgesetz:
Die Verarbeitung von Adressen auf dieser und allen unter http://www.cs.inf.ethz.ch/~rauch/ zu findenden Seiten zu Marketing- oder Werbezwecken ist untersagt.
The use of any e-mail addresses on this page and any page under http://www.cs.inf.ethz.ch/~rauch/ for marketing purposes is strictly forbidden.

[ CS-Department | Laboratory for Computer Systems | Research Group ]

Last change: 5 April 2005

rauch@inf.ethz.ch