Measurement, analysis and performance improvement of the Apache Web server
Published 1 January 1999Open access
Yiming Hu, Ashwani Nanda, Qing Yang
Citations111
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This paper measured and analyzed the behavior of the popular Apache Web server on a uniprocessor system and a 4-CPU SMP (Symmetric Multi-Processor) system running the IBM AIX operating system, and proposed and implemented 6 techniques that improve the throughput of Apache by 61%.
Abstract
Performance of Web servers is critical to the success of many corporations and organizations. However, very few results have been published that quantitatively study the server behavior and identify the performance bottlenecks. In this paper we measured and analyzed the behavior of the popular Apache Web server on a uniprocessor system and a 4-CPU SMP (Symmetric Multi-Processor) system running the IBM AIX operating system. Using the AIX built-in tracing facility, we obtained detailed information on kernel events and system activities while running Apache driven by the SPECweb96 and the WebStone benchmarks. After quantitatively identifying the performance bottlenecks, we proposed and implemented 6 techniques that improve the throughput of Apache by 61%. These techniques are general purpose and can be applied to other Web servers as well.
Keywords
Computer Science
IEEE/ACM Transactions on NetworkingSelf-similarity in World Wide Web traffic: evidence and possible causes
2,703 Citations1997Mark Crovella, Azer Bestavros
It is shown that the self-similarity in WWW traffic can be explained based on the underlying distributions of WWW document sizes, the effects of caching and user preference in file transfer, the effect of user "think time", and the superimposition of many such transfers in a local-area network.
Minds at UW (University of Wisconsin)Cost-aware WWW proxy caching algorithms
919 Citations1997Pei Cao, Sandy Irani
lmbench: portable tools for performance analysis
834 Citations1996Larry McVoy, Carl Staelin
lmbench is a micro-benchmark suite designed to focus attention on the basic building blocks of many common system applications, such as databases, simulations, software development, and networking.
IEEE/ACM Transactions on NetworkingInternet Web servers: workload characterization and performance implications
732 Citations1997Martin Arlitt, Carey Williamson
The paper concludes with a discussion of caching and performance issues, using the observed workload characteristics to suggest performance enhancements that seem promising for Internet Web servers.
A Hierarchical Internet Object Cache
691 Citations1995Anawat Chankhunthod, Peter B. Danzig +3 more
The design and performance of a hierarchical proxy-cache designed to make Internet information systems scale better are discussed, and performance measurements indicate that hierarchy does not measurably increase access latency.
The Design and implementation of the 4.3BSD UNIX operating system
616 Citations1991Marshall Kirk McKusick, Keith Bostic +2 more
Characterizing reference locality in the WWW
465 Citations2002Virgı́lio Almeida, Azer Bestavros +2 more
The authors propose models for both temporal and spatial locality of reference in streams of requests arriving at Web servers and show that temporal locality can be characterized by the marginal distribution of the stack distance trace, and proposed models for typical distributions and compare their cache performance to the traces.
Caching Proxies: Limitations and Potentials
445 Citations1995Marc Abrams, Charles R. Standridge +3 more
This work assesses the potential of proxy servers to cache documents retrieved with the HTTP, GOPHER, FTP, and WAIS protocols using World Wide Web browsers to find that with workloads over a one-semester period, a proxy has a 30--50% maximum possible hit rate.
Computer Networks and ISDN SystemsA caching relay for the World Wide Web
294 Citations1994S. Glassman
The design and performance of a caching relay for the World Wide Web is described and the data gathered from the relay is used to make some general characterizations about the web.
Self-similarity in World Wide Web traffic
262 Citations1996Mark Crovella, Azer Bestavros
It is shown that the self-similarity in WWW traffic can be explained based on the underlying distributions of WWW document sizes, the effects of caching and user preference in file transfer, the effect of user "think time", and the superimposition of many such transfers in a local area network.
Computer Networks and ISDN SystemsWorld-Wide Web proxies
228 Citations1994Ari Luotonen, Kevin Altis
An overview of proxies is given and a brand new feature is caching performed by the proxy, resulting in shorter response times after the first document fetch, making proxies useful even to the people who have full Internet access and do not really need the proxy just to get out of their local subnet.
Application-level document caching in the Internet
150 Citations2002Azer Bestavros, R.L. Carter +4 more
The results suggest that distinguishing between documents produced locally and those produced remotely can provide useful leverage in designing caching policies, because of differences in the potential for sharing these two document types among multiple users.
Measurement, analysis and performance improvement of the Apache Web server
111 Citations1999Yiming Hu, Ashwani Nanda +1 more
This paper measured and analyzed the behavior of the popular Apache Web server on a uniprocessor system and a 4-CPU SMP (Symmetric Multi-Processor) system running the IBM AIX operating system, and proposed and implemented 6 techniques that improve the throughput of Apache by 61%.
IEEE MicroVirtual-address caches. Part 1: problems and solutions in uniprocessors
96 Citations1997M. Cekleov, Michel Dubois
This survey introduces the problems and discusses solutions in the context of single-processor systems, to catalog all solutions, past and present, and to identify technology trends and attractive future approaches.
Computer Networks and ISDN SystemsMain memory caching of Web documents
81 Citations1996Evangelos P. Markatos
It is shown that even a small amount of main memory that is used as a document cache, is enough to hold more than 60% of the documents requested, and that traditional file system cache management methods are inappropriate for managing Main Memory Web caches.
Performance issues of enterprise level web proxies
77 Citations1997Carlos Maltzahn, Kathy J. Richardson +1 more
A study of two web proxies executing real workloads on Digital's Palo Alto Gateway shows that the simple CERN proxy architecture outperforms all but the latest version of Squid and continues to outperform cacheless configurations.
Measuring the Behavior of a World-Wide Web Server
77 Citations1997Jussara M. Almeida, Virgı́lio Almeida +1 more
A significant increase in resource consumption is observed along three dimensions: the number of HTTP processes running at the same time, CPU utilization, and memory utilization, which emphasizes the important role of operating system and network protocol implementation in determining Web server performance.
Server operating systems
72 Citations1996M. Frans Kaashoek, Dawson Engler +2 more
A prototype server OS is designed and implemented with support for aggressive specialization, direct device-to-device access, an event-driven organization, and dynamic compiler-assisted ILP, and an HTTP server that outperforms servers running on a conventional OS by more than an order of magnitude.
ComputerContainer shipping: operating system support for I/O-intensive applications
71 Citations1994Joseph Pasquale, Eric W. Anderson +1 more
An I/O pipeline model is presented, a model of a dynamic computation structure consisting of a sequence of domains: an input domain followed by one or more intermediate domains, and an output domain, and the design for such a facility is presented.
ACM Transactions on Computer SystemsThe measured performance of personal computer operating systems
53 Citations1996J. Bradley Chen, Yasuhiro Endo +5 more
Overall, system functionality can be accessed most efficiently in NetBSD; this is attributed to its monolithic structure and to the absence of the complications created by hardware backward-compatibility requirements in the other systems.
On the Interaction Between an Operating System and Web Server
12 Citations1997David J. Yates, Virgı́lio Almeida +1 more
This paper examines how and why web server performance changes as the workload at the server varies, using a tool that was developed, called WebMonitor, which measures activity and resource consumption, both in the operating system and in the web server.
Rice University's digital scholarship archive (Rice University)IO-LITE: a copy-free UNIX I/O system
7 Citations1997Vivek S. Pai
AIX/6000: Internals and Architecture
3 Citations1996David Kelly
How to study the AIX kernel overview of AIX, version 4 the Risc System/6000 AIX programs and processes using Crash.