This paper has benefited from the detailed and perceptive comments of our reviewers, especially our shepherd Hank Levy. We thank Randy Katz and Eric Anderson for their detailed readings of early drafts of this paper, and David Culler for his ideas on TACC's potential as a model for cluster programming. Ken Lutz and Eric Fraser configured and administered the test network on which the TranSend scaling experiments were performed. Cliff Frost of the UC Berkeley Data Communications and Networks Services group allowed us to collect traces on the Berkeley dialup IP network and has worked with us to deploy and promote TranSend within Berkeley. Undergraduate researchers Anthony Polito, Benjamin Ling, and Andrew Huang implemented various parts of TranSend's user profile database and user interface. Ian Goldberg and David Wagner helped us debug TranSend, especially through their implementation of the rewebber

System Design Issues for Internet Middleware Services: Deductions from a Large Client Trace by Steven D. Gribble Master of Science in Computer Science University of California at Berkeley Professor Eric A. Brewer, Chair In this thesis, we present the analysis of a large client-side web trace gathered from the Home IP service at the University of California at Berkeley. Specifically, we demonstrate the heterogeneity of web clients, the existence of a strong and very predictable diurnal cycle in the clients' web activity, the burstiness of clients' requests at small time scales (but not large time scales, implying a lack of self-similarity), the presence of locality of reference in the clients' requests that is a strong function of the client population size, and the high latency that services encounter when delivering data to clients, implying that services will need to maintain a very large number of simultaneously active requests. We then present system design issues for Internet midd...

The explosive growth of the Internet, the widespread use of WWW-related applications, and the increased reliance on client-server architectures places interesting new demands on network servers. In particular, the operating system running on such systems needs to manage the machine’s resources in a manner that maximizes and maintains throughput under conditions of high load. We propose and evaluate a new network subsystem architecture that provides improved fairness, stability, and increased throughput under high network load. The architecture is hardware independent and does not degrade network latency or bandwidth under normal load conditions.

The widespread use of the World Wide Web and related applications places interesting performance demands on network servers. The ability to measure theeffect of these demands is important for tuning and optimizing the various software components that make up a Web server. To measure these effects, it is necessary to generate realistic HTTP client requests. Unfortunately, accurate generation of such traffic in a testbed of limited scope is not trivial. In particular, the commonly used approach is unable to generate client request-rates that exceed the capacity of the server being tested even for short periods of time. This paper examines pitfalls that one encounters when measuring Web server capacity using a synthetic workload. We propose and evaluate a new method for Web traffic generation that can generate bursty traffic, with peak loads that exceed the capacity of the server. Finally, we use the proposed method to measure the performance of a Web server.

UNIX Internet servers with an event-driven architecture often perform poorly under real workloads, even if they perform well under laboratory benchmarking conditions. We investigated the poor performance of event-driven servers. We found that the delays typical in wide-area networks cause busy servers to manage a large number of simultaneous connections. We also observed that the select system call implementation in most UNIX kernels scales poorly with the number of connections being managed by a process. The UNIX algorithm for allocating file descriptors also scales poorly. These algorithmic problems lead directly to the poor performance of event-driven servers. We implemented scalable versions of the select system call and the descriptor allocation algorithm. This led to an improvement of up to 58% in Web proxy and Web server throughput, and dramatically improved the scalability of the system.

Enterprise level web proxies relay world-wide web traffic between private networks and the Internet. They improve security, save network bandwidth, and reduce network latency. While the performance of web proxies has been analyzed based on synthetic workloads, little is known about their performance on real workloads. In this paper we present a study of two web proxies (CERN and Squid) executing real workloads on Digital's Palo Alto Gateway. We demonstrate that the simple CERN proxy architecture outperforms all but the latest version of Squid and continues to outperform cacheless configurations. For the measured load levels the Squid proxy used at least as many CPU, memory, and disk resources as CERN, in some configurations significantly more resources. At higher load levels the resource utilization requirements will cross and Squid will be the one using fewer resources. Lastly we found that cache hit rates of around 30% had very little effect on the requests service time.

Web content hosting, in which a Web server stores and provides Web access to documents for different customers, is becoming increasingly common. Due to the variety of customers (corporate, individuals, etc.), providing differentiated levels of service is often an important issue for the hosts. Most server implementations, however, are not structured to service requests based on different levels of quality of service (QoS). This paper presents our attempts at augmenting a popular server implementation with differentiated QoS features. We explore priority-based request scheduling at both user and kernel levels. We find that simple strategies such as controlling the numbers of processes can improve the response time of high-priority requests notably while preserving the system throughput. We also find that the kernellevel approach tends to penalize low-priority requests less significantly than the user-level approach, while improving the performance of high-priority requests similarly. Ba...

The World Wide Web and its related applications place substantial performance demands on network servers. The ability to measure the effect of these demands is important for tuning and optimizing the various software components that make up a Web server. To measure these effects, it is necessary to generate realistic HTTP client requests in a test-bed environment. Unfortunately, the state-of-the-art approach for benchmarking Web servers is unable to generate client request rates that exceed the capacity of the server being tested, even for short periods of time. Moreover, it fails to model important characteristics of the wide area networks on which most servers are deployed (e.g. delay and packet loss). This paper examines pitfalls that one encounters when measuring Web server capacity using a synthetic workload. We propose and evaluate a new method for Web traffic generation that can generate bursty traffic, with peak loads that exceed the capacity of the server. Our method also mod...

This note briey summarizes some results from two papers: [4] and [23]. These papers pose the following question: Is it possible to reduce the expected response time of every request at a web server, simply by changing the order in which we schedule the requests? In [4] we approach this question analytically via an M/G/1 queue. In [23] we approach the same question via implementation involving an Apache web server running on Linux.

Widely-used operating systems provide inadequate support for large-scale Internet server applications. Their algorithms and interfaces fail to efficiently support either event-driven or multi-threaded servers. They provide poor control over the scheduling and management of machine resources, making it difficult to provide robust and controlled service. We propose new UNIX interfaces to improve scalability, and to provide fine-grained scheduling and resource management. 1 Introduction The performance of Internet server applications on a general purpose operating system is often dismayingly lower than what one would expect from the underlying hardware. Internet servers also suffer from other undesirable properties such as poor scalability, unfair resource allocation, susceptibility to livelock under excess load, instability under denial of service attacks, and inability to prioritize handling of requests. The cause of these problems is a fundamental mismatch between the original design ...