Measuring the impact of event dispatching and concurrency models on Web server performance over high-speed networks

Hu, James C.; Pyarali, Irfan; Schmidt, Douglas C.

doi:10.1109/glocom.1997.644610

Cited by 47 publications

(40 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…While the pros and cons of different server architectures have been covered extensively [15,17,12,18,8,1,25,22,9,7], we believe that insufficient attention has been devoted to the fact that a user-level thread library like Capriccio is built on an eventdriven foundation. As a result, the user-level thread-library incurs overheads that are most often only associated with eventdriven applications.…”

Section: Discussionmentioning

confidence: 99%

Comparing the performance of web server architectures

Pariag

Brecht

Harji

et al. 2007

Proceedings of the 2nd ACM SIGOPS/EuroSys European Conference on Computer Systems 2007

View full text Add to dashboard Cite

In this paper, we extensively tune and then compare the performance of web servers based on three different server architectures. The µserver utilizes an event-driven architecture, Knot uses the highly-efficient Capriccio thread library to implement a thread-perconnection model, and WatPipe uses a hybrid of events and threads to implement a pipeline-based server that is similar in spirit to a staged event-driven architecture (SEDA) server like Haboob.We describe modifications made to the Capriccio thread library to use Linux's zero-copy sendfile interface. We then introduce the SYmmetric Multi-Processor Event Driven (SYMPED) architecture in which relatively minor modifications are made to a single process event-driven (SPED) server (the µserver) to allow it to continue processing requests in the presence of blocking due to disk accesses. Finally, we describe our C++ implementation of WatPipe, which although utilizing a pipeline-based architecture, excludes the dynamic controls over event queues and thread pools used in SEDA. When comparing the performance of these three server architectures on the workload used in our study, we arrive at different conclusions than previous studies. In spite of recent improvements to threading libraries and our further improvements to Capriccio and Knot, both the event-based µserver and pipeline-based WatPipe server provide better throughput (by about 18%). We also observe that when using blocking sockets to send data to clients, the performance obtained with some architectures is quite good and in one case is noticeably better than when using non-blocking sockets.

show abstract

Section: Discussionmentioning

confidence: 99%

Comparing the performance of web server architectures

Pariag

Brecht

Harji

et al. 2007

Proceedings of the 2nd ACM SIGOPS/EuroSys European Conference on Computer Systems 2007

View full text Add to dashboard Cite

show abstract

“…Our research [20,21] demonstrates that it is possible to improve server performance through superior server design (a similar observation was made in [22]). Thus, while a "hardcoded" server, i.e., one that uses fixed concurrency, I/O, and caching strategies, can provide excellent performance, a flexible server framewor like JAWS need necessarily not perform poorly.…”

Section: Jaws Web Server Performancementioning

confidence: 67%

Developing flexible and high-performance Web servers with frameworks and patterns

Schmidt

2000

ACM Comput. Surv.

View full text Add to dashboard Cite

show abstract

“…We have refined the temporal predictability of CIAO configuration, using efficient static assembly and deployment mechanisms that are suitable for real-time platforms on which many such applications are hosted [49]. Providing fine-grain customization: In addition to the ability to configure application-level component configurations, TAO offers a variety of strategies to provide fine-grain customization of ORB mechanisms including pluggable network protocols [30]; event demultiplexing [17]; and concurrency and message dispatching [22]. These strategies allow fine-grain tuning of TAO's performance to the particular needs of each application.…”

Section: Solution Approachesmentioning

confidence: 99%

Resource virtualization in real-time CORBA middleware

Gill

2006

Proceedings of the 4th International Conference on Hardware/Software Codesign and System Synthesis

View full text Add to dashboard Cite

Middleware for parallel and distributed systems is designed to virtualize computation and communication resources so that a more abstract and consistent view of those resources is presented to the applications that use them. Providing such a consistent virtualization in distributed real-time and embedded systems becomes increasingly challenging due to application constraints such as timeliness and resource constraints such as CPU speed, power, memory, and bandwidth limitations, which also must be considered.This paper describes several examples of real-time CORBA middleware and examines how different constraints impact the way in which resources are virtualized in each case. Particular attention is paid to which details are hidden from users of the middleware, which details are exposed in the middleware's programming model, and how the hidden and exposed details interact to shape middleware design and implementation choices.

show abstract

Measuring the impact of event dispatching and concurrency models on Web server performance over high-speed networks

Abstract: Abstract

Cited by 47 publications

References 6 publications

Comparing the performance of web server architectures

Comparing the performance of web server architectures

Developing flexible and high-performance Web servers with frameworks and patterns

Resource virtualization in real-time CORBA middleware

Contact Info

Product

Resources

About