• "Jockey: user-space library for record/replay debugging" (PDF), Yasushi Saito,
  Sixth International symposium on automated and analysis-driven debugging (AADEBUG 2005), Monterey, CA, Sep 2005.
  (See below for software download).
• "Optimistic replication" (PDF), Yasushi Saito and Marc Shapiro,
  ACM Computing Surveys, Mar 2005.
• "FAB: building reliable enterprise storage systems on a shoestring" (PDF), Yasushi Saito, Svend Frolund, Alistair Veitch, Arif Merchant, Susan Spence,
  ASPLOS, Oct 2004, Boston, MA.
• "Devirtualizable virtual machines enabling general, single-node, online maintenance" (PDF), David E. Lowell, Yasushi Saito, and Eileen J. Samberg,
  ASPLOS, Oct 2004, Boston, MA.
• "A Decentralized Algorithm for Erasure-Coded Virtual Disks", (PDF) Svend Frolund, Arif Merchant, Yasushi Saito, Susan Spence, Alistair Veitch, DSN, Jun 2004, Florence, Italy.
• "FAB: enterprise storage systems on a shoestring", (PDF), Svend Frolund, Arif Merchant, Yasushi Saito, Susan Spence, and Alistair Veitch, HOTOS, May 2003, Kauai, Hawaii.
• "Taming aggressive replication in the Pangaea wide-area file system", (PDF, HTML), Yasushi Saito, Christos Karamanolis, Magnus Karlsson, and Mallik Mahalingam, 5th Symp. on Op. Sys. Design and Impl. (OSDI), Dec 2002, Boston, USA.
• "Pangaea: a symbiotic wide-area file system", (PDF), Yasushi Saito and Christos Karamanolis, to appear in ACM SIGOPS European Workshop, Sep 2002, Saint Emilion (near Bordeax), France.
• "Optimistic replication for Internet data services" (fixed kinks on 10/26) (PDF(two-column treesaver), PS (LNCS format), Slides)
  Yasushi Saito and Henry Levy, 14th International Conference on Distributed Computing (DISC), Oct 2000, Toledo, Spain.
  Also appears in LNCS 1914, pp297-314.
  This paper discusses the replication algorithm used by Porcupine. Complete with correctness proof. By the way, too many people confuse DISC and ICDCS. DISC is a European distributed theory conference, whereas ICDCS is a kitchen sink distributed systems conference.
• "Manageability, Availability and Performance in Porcupine: A Highly Scalable Internet Mail Service"(PS)
  Yasushi Saito, Brian Bershad, and Henry Levy. ACM Transactions on Computer Systems, Aug 2000.
  This paper is an extended version of the SOSP paper.
• "Manageability, Availability and Performance in Porcupine: A Highly Scalable Internet Mail Service" (PS, Slides)
  Yasushi Saito, Brian Bershad, and Hank Levy. 17th ACM Symposium on Operating Systems Principles, Dec 1999, Kiawah Island Resort, near Charleston, SC, USA.
  Best paper award
• "Application performance pitfall and TCP's Nagle algorithm", (PS)
  Greg Minshall, Yasushi Saito, Jeffrey Mogul, and Ben Verghese, Workshop on Internet Server Performance, May 1999, Atlanta, GA, USA.
• "The Porcupine Scalable Mail Server" (PS, Word, Slides)
  Yasushi Saito, Eric Hoffman, Brian Bershad, Henry Levy, David Becker and Bertil Folliot. ACM SIGOPS European Workshop, Sep 1998, Sintra, Portugal.
  This paper overviews the Porcupine mail server, a high performance mail server we are developing.
• "Transaction Support in an Extensible Operating System"(PS)
  Yasushi Saito and Brian Bershad, USENIX Annual Technical Conference, Jun 1998, New Orleans, LA, USA.
  This paper describes a transactionsal memory implemented in SPIN. Three mechanisms are compared using two benchmarks, RVMbench and OO7.

Talks and presentations

• FAB talk at Stanford (5/27/2003) and CMU (6/24/2003) (sxi, PDF)
• FAB OSDI WIP talk slides (12/11/2002) (sxi)
• Pangaea OSDI talk slides (12/11/2002) (sxi)
• Pangaea SOSP-WIP talk slides (10/24/2001) (ppt).
• Pangaea 1hr talk slides (last updated: 09/30/2002) (sxi). These slides were used in talks at HPL CSC (June 2002) and MSR Cambridge (Sep 2002).
• Job talk slides (ppt).
• What happened since your quals?, a.k.a. an extremely biased view of recent operating system research I did this informal talk at a UW CS grad meeting in Jan, 2000.

Non-refereed publications

• "Jockey: a user-space library for record-replay debugging" (pdf), Yasushi Saito, HPL TR 2005-46
• "RAMBO for dummies" (pdf), Jeannie Albrecht and Yasushi Saito, HPL TR 2005-39.
• "A decentralized algorithm for erasure-coded virtual disks", Svend Frolund, Yasushi Saito, Arif Merchant, Susan Spence, and Alistair Veitch. HPL-SSP-2003-14 (this is a longer version of the DSN paper), Oct 2003.
• "Scaling optimistic replication algorithms", Marc Shapiro and Yasushi Saito, a position paper appeared in the workshop on future directions in distributed computing (FuDiCo), Bertinolo, Italy, June, 2002, (LNCS whatever) (pdf).
• "Name space consistency in the Pangaea wide-area file system", Saito and Karamanolis. (pdf), HPL SSP Technical Memo HPL-SSP-2002-12. .
  This paper discusses the replication protocol used in Pangaea in gory detail.
• "Unilateral version vector pruning using loosely synchronized clocks", (pdf), Technical report HPL-2002-??.
• "Decentralized and autonomous replication in the Pangaea planetary-scale file service", (pdf), HPL-TR-2001-323.
• "Functionally homogeneous clustering: a framework for building scalable data-intensive Internet services", Ph.D. dissertation, University of Washington, June 2001. (pdf)
• "A Usenet Performance Study" (ps)
  Yasushi Saito, Jeff Mogul, and Ben Verghese, Sep 1998. (a project report while at Compaq WRL, which is now a part of HP Labs.)

Deprecated publications

• "Optimistic replication", Yasushi Saito and Marc Shapiro (pdf), Technical report MSR-TR-2003-60, Oct 2003.
  This techreport is superceded by the ACM-CS submission (see above)
• "Replication: Optimistic Approaches", Yasushi Saito and Marc Shapiro (at MSR Cambridge) (pdf), Technical report HPL-2002-33.
  This techreport is superceded by ACM-CS submission.
• "Consistency management in optimistic replication algorithms" (pdf). My general exam report. A survey of OR algorithms.
  This report is superceded by the ACM-CS submission

• Jockey execution record/replay library (new! Oct 21, 2005)
• Disk I/O trace logger for linux 2.6
• Linux vectored aio patch
  This is a kernel patch that adds readv and writev to libaio.
• oopsfrag This horrendous hack allows any LaTeX string to be spliced into an OpenOffice picture. Sort of like psfrag, but it's specialized for OpenOffice. It transforms OO Postscript files directly, so it should work with pdflatex as well.
• Andrew-tcl benchmark. This is a variation of the infamous Andrew benchmark that tests a file system's performance. This one uses Tcl source code instead of some arcane X application.
• pstopng. Convert EPS to PNG. Requires the newest ghostscript and perl.
• PyChart, a Python-based scientific-chart drawing tool.
• Octopy, a Qt frontend for Arch revision control system.
• pprof, Multi-threaded profiling support for Linux new! (2/5/2001).
• Porcupine, an experimental cluster-based mail server.(1/19/2001)
• Newsbench, Usenet traffic generator and server benchmarking suite. See this report for the background.
• Command-line Concise Oxford dictionary client and Merrian-Webster Collgiate dictionary client. They require Perl. new!(12/23/2000)