Recent Papers

High-Performance Local Area Communication With Fast Sockets, Steve H. Rodriguez, Thomas E. Anderson, Davie E. Culler, To Appear in Usenix'97.

Modern switched networks such as ATM and Myrinet enable low-latency, high-bandwidth communication. This performance has not been realized by current applications, because of the high processing overheads imposed by existing communications software. These overheads are usually not hidden with large packets; most network traffic is small. We have developed Fast Sockets, a local-area communication layer that utilizes a high-performance protocol and exports the Berkeley Sockets programming interface. Fast Sockets realizes round-trip transfer times of 60 microseconds and maximum transfer bandwidth of 33 MB/second between two UltraSPARC 1s connected by a Myrinet network. Fast Sockets obtains performance by collapsing protocol layers, using simple buffer management strategies, and utilizing knowledge of packet destinations for direct transfer into user buffers. Using receive posting, we make the Sockets API a single-copy communications layer and enable regular Sockets programs to exploit the perfomance of modern networks. Fast Sockets transparently reverts to standard TCP/IP protocols for wide-area communication.

Effects of Communication Latency, Overhead, and Bandwidth in a Cluster Architecture, Richard P. Martin, Amin M. Vahdat, David E. Culler, and Thomas E. Anderson, Submitted for Publication.

This work privides a systematic study of the impact of communication performance on parallel applications in a high performance cluster of workstations. We develop an experimental system in which the communication latency, overhead, and bandwidth can be independently varied to observe the effects on applications demonstrating a wide range of architectural requirements. Our results indicate that efforts to bring cluster communication performance in line with the performance of more tightly integrated parallel machines have resulted in significantly improved application performance. We show that applications demonstrate strong sensitivity to overhead and message bandwith, slowing down by up to a factor of 30 on sixteen processors when overhead is increased by 100 us. Surprisingly, many of our benchmark applications are tolerant of increased latency and lower bulk message bandwidth. Finally, applications demonstrate a linear dependence to both overhead and gap, indicating that further improvements in communication performance will continue to improve application performance.

High-Performance Sorting on Networks of Workstations, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-Dusseau, Joseph M. Hellerstein, and David A. Patterson, Submitted for Publication.

We report the performance of NOW-Sort, a collection of sorting implementations on a Network of Workstations (NOW). We find that parallel sorting on NOWs is competitive to sorting on the large-scale SMPs that have traditionally held the performance records. On a 32-node cluster, we finish the Datamation benchmark in 2.41 seconds, and can sort 3.0 GB in just under one minute. On a smaller, better equipped, 8-node cluster, we run the Datamation in 2.92 seconds,and sort 1.4 GB in a minute.
Our implementations can be applied to a variety of disk, memory, and processor configurations; we highlight salient issues for tuning each component of the system. Throughout the paper, we evaluate the use of commodity hardware and operating systems for parallel sorting, and note lessons that can be drawn when applying NOW technology to data-intensive applications.

Evaluation of Architectural Support for Global Address-Based Communication in Large-Scale Parallel Machines, Arvind Krishnamurthy, Klaus E. Schauser, Chris J. Scheiman, Randolph Y.wang, David E. Culler, and Katherine Yelick, Appeared in ASPLOS'96.

Using Smart Clients to Build Scalable Services, Chad Yoshikawa, Brent Chun, Paul Eastham, Amin Vahdat, Thomas Anderson, and David Culler, To Appear in Usenix'97.

Individual machines are no longer sufficient to handle offered load to many Internet sites. To use multiple machines for scalable performance, issues involving load balancing, fault tolerance, and backward compatibility with URL naming must be addressed . A number of server-side approaches have been developed to provide this level of transparent access to Internet services, including HTTP redirect, DNS aliasing, Magic Routers, and Active Networks. In this paper, we will argue that in many instances, a client-side approach to providing transparent access to Internet services provides increased flexibility and performance. We describe the design and implementation of {\em Smart Clients} and show how our system can be used to provide a transparently scalable and highly available interface to three network services: telnet, ftp, and an Internet chat application.

WebFS: A Global Cache Coherent File System, Amin Vahdat, Paul Eastham, and Thomas Anderson, Submitted for Publication.

The immense popularity of the World Wide Web has lead to the development of a vast number of new distributed applications. These distributed Internet applications often require common underlying functionality, including: a global name space, cache coherence, fault tolerance, security, and authentication. Unfortunately, this functionality must currently be reimplemented on a per-application basis. We are building WebFS, a global cache coherent file system to provide a common substrate for developing Internet applications. A prototype of WebFS has been completed for the Solaris operating system and provides access to the HTTP name space for unmodified binaries. A number of cache coherence policies have also been implemented providing support for two prototype applications, scalable Internet chat and a stock ticker. In particular, we have found the use of multicast updates appropriate for applications which broadcast regular updates to a dynamic, global audience.

WebOS: Software Support for Scalable Web Services, Thomas E. Anderson, David E. Culler, NSF Proposal.


Previous                    Home

Andrea Arpaci-Dusseau and Randy Wang ({dusseau,rywang}@cs.berkeley.edu )

This page created with Netscape Navigator Gold (last updated on December 22, 1996)