MIME-Version: 1.0
Server: CERN/3.0
Date: Sunday, 24-Nov-96 23:03:44 GMT
Content-Type: text/html
Content-Length: 5002
Last-Modified: Monday, 20-May-96 20:03:27 GMT
Werner Vogels
|
Werner Vogels
researcher
4105a Upson Hall
Dept. of Computer Science
Cornell University,
Ithaca, NY, 14850
|
Phone: 607-255-9199
Fax : 607-255-4428
Email: vogels@cs.cornell.edu
|
In protocol design, perfection has been reached not when there is nothing
left to add, but when there is nothing left to take away.
I am involved with two of the major systems projects at the Cornell Computer Science Department: The
Horus
and the
Cornell ATM Cluster Projects. I think my
research interests are best described by:
Low latency and high bandwith communication support for
highly reliable distributed systems with real time requirments.
I focus on the system design and engineering issues. Some of the things I am
working on:
-
A highly predictable execution environment for Horus. The integration of Horus
with some Real-Time environments needs to lead to a situation where we can
reason about advanced operational guarantees.
-
Mechanisms for efficient data transfers between high-speed
network devices and the application level. Low latency for all messages
and high bandwidth with small messages are two issues that
have fallen behind in the software design for high-speed network adapters.
-
High-speed Cluster Communication protocols. Once you achieve the desired low-latency
for your message passing system, you will see that your old protocols
are not able to exploit this, and you will need to re-think their structure
and interaction patterns.
-
Methods for dealing with guarantee failures (aka
missed deadline support). It is not bad to not be able to meet the
guarantes you gave, it is bad to not tell anyone about it.
-
Acurate Failure Detection. If we want to take our distributed systems to a
global scope, there is a need to find a generic mechanism to support
failure suspision, detection and management of processes, nodes and networks.
From our experience with group systems we can extract a mechanism that will work
with any middleware package, regardless of its functionality.
Horus is the brainchild of
Robbert van Renesse and
Ken Birman. The ATM Cluster work is done cooperation with
Thorsten von Eicken and the Multimedia & Video-On-Demand Horus
experiments are in concert with
Brian Smith.
I am responsible for
CS 515,
a Practicum in Distributed Systems, and teach a number of lectures
on network protocols and high-speed network technology in
CS 514:
Practical Distributed Systems.
Recent publications:
-
World Wide Failures, Werner Vogels,
To appear in the Proceeding of the 1996 ACM SIGOPS Workshop
Connamoran, Ierland,
September 1996.
-
Structured Virtual Synchrony: Exploring the Bounds of Virtually
Synchronous Group Communication.
Katherine Guo, Werner Vogels, Robbert van Renesse,
To appear in the Proceeding of the 1996 ACM SIGOPS Workshop
Connamoran, Ierland,
September 1996.
-
U-Net: A User-Level Network Interface
for Parallel and Distributed Computing,
Anindya Basu, Vineet Buch, Werner Vogels, Thorsten von Eicken,
Proceedings of the 15th ACM Symposium on Operating
Systems Princples, Copper Mountain, December 1995.
-
Delivering High-Performance Communication to the Application-Level.
Werner Vogels and Thorsten von Eicken, in the Proceeding of
the Third IEEE Workshop on the Architecture and Implementation
of High Performance Communication Subsystems (HPCS'95), August 1995.
-
Horus: A Flexible Group Communications System ,
Robbert van Renesse, Kenneth P. Birman, Brad Glade, Katie Guo, Mark
Hayden, Takako Hickey, Dalia Malki, Alex Vaysburd and Werner Vogels,
CS-TR 95-1500, March 23, 1995.