http:^^simon.cs.cornell.edu^info^projects^horus^abstracts.html
来自「This data set contains WWW-pages collect」· HTML 代码 · 共 520 行 · 第 1/2 页
HTML
520 行
MIME-Version: 1.0
Server: CERN/3.0
Date: Monday, 16-Dec-96 21:45:06 GMT
Content-Type: text/html
Content-Length: 23607
Last-Modified: Tuesday, 19-Nov-96 20:35:33 GMT
<HTML><HEAD><TITLE>Abstracts of the Horus Papers</TITLE></HEAD><BODY BACKGROUND=images/per3.jpg TEXT=#00000><TABLE WIDTH=800 BORDER=0><TR><TD WIDTH=140 VALIGN=TOP browspan=1ALIGN=LEFT><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/ud8.jpg"><BR><P><BR><P><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/index.html"><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-home.jpg"></A><BR><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/Overview.html"><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><!WA4><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-overview.jpg"></A><BR><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><!WA5><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/People.html"><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><!WA6><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-people.jpg"></A><BR><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><!WA7><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/Papers.html"><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><!WA8><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-papers.jpg"></A><BR><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><!WA9><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/Software.html"><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><!WA10><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-software.jpg"></A><BR><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><!WA11><A HREF="http://simon.cs.cornell.edu/Info/Projects/HORUS/Links.html"><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><!WA12><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-links.jpg"></A><BR><P><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><!WA13><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/recent2.jpg"><P><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><!WA14><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/sciam2.jpg"><P><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><!WA15><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/sigcomm.jpg"><P><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><!WA16><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/cacm2.jpg"><P><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><!WA17><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/tina.jpg"><P><BR><P><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><!WA18><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/ensicon2.gif"><P><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><!WA19><A HREF="http://simon.cs.cornell.edu/Info/Projects/Ensemble/index.html"><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><!WA20><IMG BORDER=0 SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/button-ens-home.jpg"></A><P></TD><TD WIDTH=640><CENTER><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><!WA21><IMG SRC="http://simon.cs.cornell.edu/Info/Projects/HORUS/images/papers.jpg"><BR></CENTER><CENTER><H2>Abstracts</H2><P></CENTER><HR><A NAME="sciam"></A><H2>Software for Reliable Networks</H2>Kenneth P. Birman and Robbert van Renesse<BR>Scientific American, May, 1996<BR><P>The failure of a single program on a single computer can sometimes crash a network of intercommunicating machines, causing havoc for stock exchanges, telephone systems, air-traffic control and other operations. Two softwaredesigners explain what can be done to make networks more robust. <P><HR><P><A NAME="cacm"></A><H2>Horus, a flexible Group Communication System</H2>Robbert van Renesse, Kenneth P. Birman and Silvano Maffeis<BR>Communications of the ACM, April 1996.<P>The emergence of process-group environments for distributed computingrepresents a promising step towards robustness for mission-critical distributed applications.Process groups have a ``natural'' correspondence with data or services that have been replicatedfor availability, or as part of a coherent cache.They can been used to support highly available securitydomains. And, group mechanisms fit well with an emerging generation ofintelligent network and collaborative work applications.<P><HR><P><A NAME="hpa"></A><H2>Masking the Overhead of Protocol Layering</H2>Robbert van Renesse<BR>Proceedings of the 1996 ACM SIGCOMM Conference<BR>Stanford, September 1996<BR><P>Layering of protocols has been advocated as a way of dealing with thecomplexity of computer communication.It has also been criticized for its performance overhead.In this paper, we present some insights in the design of protocols, andhow these insights can be used to mask the overhead of layering, in away similar to client caching in a file system.With our techniques, we achieve an order of magnitude improvement inend-to-end message latency in the Horus communication framework.Over an ATM network, we are able to send and deliver messages of varyinglevels of semantics in about 85 microseconds, using a protocol stack offour layers that were written in ML, a high-level functional language.<P><!--<HR><P><A NAME="rtss7"></A><H2>Using Group Communication Technology to Implement a Reliable andScalable Distributed IN Coprocessor</H2>Roy Friedman and Ken Birman<BR>September 1996<BR><P>In this paper we explore the use of group communication technology,developed in the Horus project, to implement a reliable and scalabledistributed IN coprocessor.The proposed implementation can handle up to 20,000 calls per secondwith 12 computing nodes, can tolerate a single node failure or recovery,and can recover from periods of overload.<P>--><HR><A NAME="worldwidefailures"></A><P><H2>World Wide Failures</H2>Werner Vogels<BR>Proceedings of the ACM SIGOPS European Workshop, Connamoran, Ireland,September 1996 <BR><P>The one issue that unites almost all approaches to distributed computingis the need to know whether certain components in the system have failed orare otherwise unavailable. When designing and building systems thatwill need to function at a global scale, failure management will need tobe considered a fundamental building block. This paper describes the development of a system-independent failure management servcies, which allowssystems and applications to incorporate accurate detection of failedprocesses, nodes and networks without the need for making compromises in theirparticular design.<P><HR><A NAME="framework"></A><P><H2>A Framework for Protocol Composition in Horus</H2><P>Robbert van Renesse, Kenneth P. Birman, Roy Friedman, <BR>Mark Hayden, and David A. Karr<BR>August 1995<P>The Horus system supports a communication architecture that treatsprotocols as instances of an abstract data type.This approach encourages developers topartition complex protocols into simple microprotocols, each of whichis implemented by a protocol layer.Protocol layers can be stacked on top of eachother in a variety of ways, at run-time.First, we describe the classes of protocols that can be supported this way.Next, we present the Horus object model that we designed forthis technology, and the interface between the layers that makes it allwork. We then present an example layer that implements a groupmembership protocol. Next, we show how, given a set of required properties,an appropriate stack can be constructed. We look at an example stack ofprotocols, which provides fault-tolerant, totally ordered communicationbetween a group of processes.The work contributes a standard framework for protocol development andexperimentation, provides a high performance implementation of thevirtual synchrony model, and introduces a methodology for increasing therobustness of the protocol development process.<P><HR><A NAME="tr95-1579"><H2>Trading Consistency for Availability in Distributed Systems</H2>Roy Friedman and Ken Birman<BR>TR96-1579<BR>April 8, 1996<BR><p>This paper shows that two important classes of actions, non left commuting and strongly non commuting, cannot be executed by concurrent partitions in a system that provides serializable services. This result indicates that there is an inherent limitation to the ability of systems to provide services in a consistent manner during network partitions.<P><HR><A NAME="tr95-1554"></A><H2>Deciding in Partitionable Networks</H2>Roy Friedman, Idit Keidar, Dalia Malki, Ken Birman and Danny Dolev<BR>TR95-1554<BR>November 27, 1995<BR><p>Motivated by Chandra and Toueg's work, we study decision protocols in a model that closely approximates ``real'' distributed systems. Our results show how the weakest failure detector and associated consensus algorithm can be adapted to a network in which omission failures can occur during periods when processes suspect one-another as faulty. For protocols in which a majority subset of the participants can reach decisions on behalf of the system as a whole, we also characterize a series of stages that necessarily arise during execution. Jointly, these findings establish a direct relationship between an extended version of the three-phase commit protocol, which makes progress even when a traditional three-phase commit would block, and the consensus protocol of Chandra and Toueg. Although we do not explore the linkage here, our results should also be applicable to other agreement protocols for systems of this sort, such as leader election and dynamic group membership.<P><HR><P><A NAME="tr95-1537"></A><P><H2>Strong and Weak Virtual Synchrony in Horus</H2> Roy Friedman and Robbert van Renesse<br> August 24, 1995 <p> A formal definition of {\em strong virtual synchrony}, capturing the semantics of virtual synchrony as implemented in Horus, is presented. This definition has the nice property that every message is delivered within the view in which it was sent. However, it is shown that in order to implement strong virtual synchrony, the application program has to block messages during view changes. An alternative definition, called {\em weak virtual synchrony}, which can be implemented without blocking messages, is then presented. This definition still guarantees that messages will be delivered within the view in which they were sent, only that it uses a slightly weaker notion of what the view in which a message was sent is. An implementation of weak virtual synchrony that does not block messages during view changes is developed, and it is shown how to use a system that provides weak virtual synchrony even when strong virtual synchrony is actually needed. To capture additional ordering requirements, the definition of {\em ordered virtual synchrony} is presented. Finally, it is discussed how to extend the definitions in order to cope with the fact that a process can become a member of more than one group.<P><hr><A NAME="tr95-1527"></A><P><H2>Packing Messages as a Tool for Boosting the Performance of Total Ordering Protocols </H2> Roy Friedman and Robbert van Renesse<br> July 07, 1995 <p> This paper compares the throughput and latency of four protocols that provide total ordering. Two of these protocols are measured with and without message packing. We used a technique that buffers application messages for a short period of time before sending them, so more messages are packed together. The main conclusion of this comparison is that message packing influences the performance of total ordering protocols under high load overwhelmingly more than any other optimization that was checked in this paper, both in terms of throughput and latency. This improved performance is attributed to the fact that packing messages reduces the header overhead for messages, the contention on the network, and the load on the receiving CPUs.<P><hr><p><A NAME="tr95-1506"></A><H2>Using Virtual Synchrony to Develop Efficient Fault TolerantDistributed Shared Memories </H2> Roy Friedman<br> March 31, 1995 <p> This paper shows how to define consistency conditions for distributed shared memories in virtually synchronous environments. Such definitions allow to develop fault tolerant implementations of distributed shared memories, in which during normal execution, operations can be performed very efficiently, and only those operations which take place during a configuration change must be delayed. Three well known consistency conditions, namely, linearizability, sequential consistency, and causal memory, are redefined for virtually synchronous environments. It is then shown how to provide efficient fault tolerant implementations for these⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?