Homepage of Peter Boncz

notable (recent) activities:

Won the VLDB 10-Year Best Paper Award together with Stefan Manegold and Martin Kersten, for our work on architecture-conscious query processing in column stores! VLDB Keynote
At VLDB 2009 I presented a tutorial on Column oriented Database Systems (PDF) together with Daniel Abadi (Yale) and Stavros Harizopoulos (HP Labs).
Together with prof. Divy Agrawal (UCSB), taught at the VLDB Summer School, held this year Aug 10-14 in Shanghai. Speaking for 2.5 days non-stop to 100 Chinese DB PhD students turned out to be.. quite intense (sheets: Wed Thu 1 Thu 2 Thu 3 Fri).
Co-founded a new company, VectorWise, around the X100 high-performance database system. More news here: we are collaborating with open-source database company Ingres.
With Ken Ross (Columbia University) I am back as organizer of DaMoN 2009, a workshop which I co-founded in 2005.
My NWO Open project proposal Querying while Transforming Large Graph Databases was awarded; the project will run 2010-2014.
MonetDB BV was founded, a company that provides services around the MonetDB open-source database system (topic of my PhD research).
Published an invited "research highlight" article in the Communications of the ACM journal, on the architecture-consious research performed in the past decade in the context of the MonetDB project (co-authors Martin Kersten and Stefan Manegold -- both from CWI).
On invitation from prof. Hector Garcia-Molina and dr. Meichun Hsu, I spoke at the 2008 InfoSeminar at Stanford University.
I won the ICT Regie Award 2006 for my role in the CWI spin-off company Data Distilleries (see below).
member of the CWI ondernemingsraad.

The INS1 research theme at CWI, where I work aims at researching database technology, applied to query-intensive domains like data mining, XML databases and multimedia retrieval.

Professional Activities

My CV is publicly available via:

Being a database architect, my research tends to be clustered around efforts to build database systems:

MonetDB/XQuery: a high-performance XML database system. It uses the Pathfinder XQuery-to-Relational Algebra compiler and (loop-lifted) staircase join algorithms to turn the binary relational MonetDB into a full-fledged XML DBMS.
AmbientDB: a P2P database architecture, supporting ad-hoc distributed querying, schema integration and data synchronization. We see AmbientDB as a "data management" enhanced middleware technology, that eases construction of intelligent applications on networks of pervasive computing devices. The query processing core of AmbientDB is MonetDB/XQuery. It uses SOAP for distributed querying, and P2P data structures for node discovery and connectivity.
X100: continues our research into the interaction between database architecture and modern computer architecture (cache-conscious processing, CPU efficiency of database algorithms, efficient exploitation sequential I/O). As this is about performance, we focus on the (few?) application domains where this is relevant. This includes on-line data mining and OLAP on large datawarehouses such as in TPC-H, but also content-based information access in huge multimedia databases like TREC Video.

All these research activities are funded as part of the 6-year project MultimediaN, by the Dutch Government (BSIK), in which I lead the database sub-project with involvement from Philips Research and Technical University Twente (CTIT). In this context, CWI has research collaborations with NFI (MonetDB/XQuery), SPSS (X100) and TextKernel BV (X100).

Ph.D. Students

These students worked in MultimediaN (2004-2009), in which I was the lead in of of the projects (14FTE at CWI, Philips Research, U Twente). I am also advising:

Eleftherios Sidirourgos
Nan Tang (post-doc)

Both positions are in the XIRAF project 2007-2009, jointly with the Dutch Forensic Institute (NFI).

PC Memberships

Major conferences:

SIGMOD 2004, SIGMOD 2008 (Demo), SIGMOD 2010
VLDB 2001,2005,2007,2008,2009,2010
ICDE 2005,2008,2009,2010
EDBT 2002,2008
WWW 2009
CIKM 2008 (area chair)

other PCs:

SMDB 2010
W3C Workshop on RDF Access to Relational Databases
XIME-P 2007
MDM 2007
ExpDB 2006
DBISP2P 2004,2005,2006,2007
VLDB PhD 2005,2006
DAMON 2007,2008
PDMST 2006,2007

Additionally, I have served as referee for major journal publications:

ACM TODS
IEEE TKDE
VLDB Journal

Organization

Co-organizer of the fifth DaMoN, the international workshop on Data Management on New Hardware, co-located with and sponsored by ACM SIGMOD 2009.
Co-organizer of the workshop on XQuery Implementation Paradigms, to be held at Schloss Dagstuhl, 19-22 november 2006.
Co-organizer of the second DaMoN, the international workshop on Data Management on New Hardware, co-located with and sponsored by ACM SIGMOD 2006.
Co-organizer of the DBDBD 2005, the Dutch-Belgian Database Day.
Co-organizer of the first DaMoN, the international workshop on Data Management on New Hardware, co-located with and sponsored by SIGMOD 2005.

Past Work

In may 2002, I completed and defended my PhD thesis on MonetDB. MonetDB is an experiment in database architecture (gone out of hand) that proposes to use full vertical fragmentation in order to better accommodate query-intensive access patterns, both in terms of I/O optimization as well as for improving the access to the memory caches. MonetDB also uses a column-wise query-processing algebra that has a zero degree of freedom, which makes it possible to use a generic but pre-compiled query engine, as opposed to the interpretative techniques used in other DBMSs. Compile-time fixed query processing primitives are crucial for modern CPUs like the Pentium 4, which need highly predictable code in order to avoid branch mispredictions, as well as an ever present pool of independent instructions (in order to fill its parallel units and obtain a good Instructions-Per-Cycle ratio). Note that compile-time here means DBMS build time, and not query-compilation-time, which is a run-time activity. A final MonetDB design issue was extensibility: by constructing MonetDB as a back-end, on top of which multiple front-end systems can work and interact with the storage/query on a lower level than, say, SQL allows to re-use the same system in multiple application domains, which was one of MonetDB's design goals. MonetDB has been applied successfully to OLAP, data mining, GIS, k-NN search, XML-, image- and video-databases.

During my time as PhD candidate at the University of Amsterdam, our research group founded Data Distilleries (DD -- it has been acquired in 2004 by SPSS), and I was with it almost from the beginning. Its decison management tools are powered by MonetDB, and as such I got deeply involved, spending more than two years full time as overall product architect at the company. As such I architected and designed large parts of the new real time suite as well, getting acquainted with J2EE application server technology in the area of real-time marketing and also real-time heterogeneous databases. As any start-up company job, this was not about software architecture alone, and I learned a lot about software engineering methodology in practice, but also on project and product management. However, the most fun part was always going on site visits to the clients of DD, which are mostly big companies (like ABN Amro, ING Postbank, Aegon, Vodafone, Center Parcs, OHRA, IMP) getting to the highly-protected machine room, where between many huge mainframes, a Unix or NT server would be (and is still) running MonetDB!

Research Interests

So, since the start of 2002 I'm back at CWI doing research into:

database architecture: the art of crafting DBMS software.
data mining, OLAP: the so-called "query-intensive" DBMS application areas.
(XML) query processing: algorithms and data (index) structures for high query performance.
non-standard application domains: applying DBMS technology to multimedia, GIS and biology.
computer architecture: the interaction between a DBMS and modern hardware.
P2P technologies: architectures that allow for ad-hoc cooperation.

Publications

CWI DISCLAIMER