José Carlos Brustoloni
I recently received my Ph.D. degree from the
School of Computer Science
at Carnegie Mellon University .
My advisor was
Peter Steenkiste , and the other thesis
committee members were David Johnson, M. Satyanarayanan, and
Most of my research at CMU targeted efficient end-to-end communication
over high speed networks. Aspects of this work included:
This research resulted in
new techniques for reducing the cost and volume
of I/O data passing between end applications and kernel- or user-level
To demonstrate and evaluate my techniques, I implemented Genie, an I/O
framework that allows applications to select the semantics of their
I/O buffers and to download I/O programs into the kernel.
My experiments on Genie demonstrated for the first time that
APIs with copy semantics (such as those of Unix and Windows NT) can
be optimized so as to have performance competitive with that of APIs with share or
move semantics, and that user-level protocol servers can be optimized so as to give
end-to-end performance approaching that of kernel-level ones.
- Virtual memory optimizations for
high-performance I/O (streamlining the data path between
applications and the network);
- Network adapter design
(hardware support for efficient data passing to and from applications);
- Operating system organization (flexible and efficient infra-structure
for protocol stack implementation and application-specific I/O
- Communications protocols (in particular, ATM flow control).
The following papers describe my research at CMU:
José Carlos Brustoloni's research / SCS-CMU /
User-Level Protocol Servers with Kernel-Level Performance, manuscript accepted for publication in
in Proceedings of the INFOCOM'98 Conference ,
IEEE, San Francisco, CA, March 1998.
Effects of Data Passing Semantics and Operating System Structure on Network I/O Performance,
Ph.D. Dissertation, Technical Report CMU-CS-97-176, School of Computer Science, Carnegie Mellon
University, September 1997.
Evaluation of Data Passing and Scheduling Avoidance,
in Proceedings of the 7th International Workshop on
Network and Operating System Support for Digital Audio and Video
(NOSSDAV'97), IEEE, St. Louis, MO, May 1997, pp. 101-111
Scaling of end-to-end latency with network transmission rate,
in Proceedings of the Gigabit Networking Workshop (GBN'97), IEEE,
Kobe, Japan, April 1997.
Copy Emulation in Checksummed, Multiple-Packet Communication,
in Proceedings of the INFOCOM'97 Conference ,
IEEE, Kobe, Japan, April 1997
Effects of Buffering Semantics on I/O Performance,
in Proceedings of the Second Symposium on
Operating Systems Design and Implementation (OSDI'96),
USENIX, Seattle, WA, October 1996, pp. 277-291
Also appears in Operating Systems Review , SIGOPS, ACM, vol. 30,
Special Issue, 1996, pp. 277-291.
Application-Allocated I/O Buffering with System-Allocated Performance,
Technical Report CMU-CS-96-169, School of Computer Science, Carnegie Mellon
University, August 1996.
Extended version of the OSDI'96 paper, "Effects of Buffering Semantics
on I/O Performance".
Exposed Buffering and Sub-Datagram Flow Control for ATM LANs,
in Proceedings of the 19th Conference on Local Computer Networks ,
IEEE, Minneapolis, MN, October 1994, pp. 324-334
Simple Protocol Processing for High-Bandwidth Low-Latency Networking,
Technical Report CMU-CS-93-132, School of Computer Science, Carnegie Mellon
University, March 1993.
Describes an early implementation of native, integrated-layer protocols
for an ATM LAN.
Autonomous Agents: Characterization and Requirements,
Technical Report CMU-CS-91-204, School of Computer Science, Carnegie Mellon
University, November 1991.
Describes a hierarchical architecture for intelligent agents, encompassing
reaction, planning, and learning. Work performed for a minor in AI.