General-purpose operating systems provide inadequate support for resource management in large-scale servers. Applications lack sufficient control over scheduling and management of machine resources, which makes it difficult to enforce priority policies, and to provide robust and controlled service. There is a fundamental mismatch between the original design assumptions underlying the resource management mechanisms of current general-purpose operating systems, and the behavior of modern server applications. In particular, the operating system's notions of protection domain and resource principal coincide in the process abstraction. This coincidence prevents a process that manages large numbers of network connections, for example, from properly allocating system resources among those connections. We propose and evaluate a new operating system abstraction called a resource container, which separates the notion of a protection domain from that of a resource principal. Resource containers ...

In this paper we present two protocols for asynchronous Byzantine Quorum Systems (BQS) built on top of reliable channels---one for self-verifying data and the other for any data. Our protocols tolerate Byzantine failures with fewer servers than existing solutions by eliminating nonessential work in the write protocol and by using read and write quorums of different sizes. Since engineering a reliable network layer on an unreliable network is difficult, two other possibilities must be explored. The first is to strengthen the model by allowing synchronous networks that use time-outs to identify failed links or machines. We consider running synchronous and asynchronous Byzantine Quorum protocols over synchronous networks and conclude that, surprisingly, "self-timing" asynchronous Byzantine protocols may offer significant advantages for many synchronous networks when network time-outs are long. We show how to extend an existing Byzantine Quorum protocol to eliminate its dependency on reliable networking and to handle message loss and retransmission explicitly.

This paper presents the design of a new Web server architecture called the asymmetric multiprocess event-driven (AMPED) architecture, and evaluates the performance of an implementation of this architecture, the Flash Web server. The Flash Web server combines the high performance of single-process event-driven servers on cached workloads with the performance of multi-process and multithreaded servers on disk-bound workloads. Furthermore, the Flash Web server is easily portable since it achieves these results using facilities available in all modern operating systems. The performance of different Web server architectures is evaluated in the context of a single implementation in order to quantify the impact of a server's concurrency architecture on its performance. Furthermore, the performance of Flash is compared with two widely-used Web servers, Apache and Zeus. Results indicate that Flash can match or exceed the performance of existing Web servers by up to 50 % across a wide range of real workloads. We also present results that show the contribution of various optimizations embedded in Flash.

Abstract In this paper, we examine several distributed caching strategies to improve the response time for accessing data over theInternet. By studying several Internet caches and workloads, we derive four basic design principles for large scale distributed

In this paper, we describe the design and implementation of an integrated architecture for cache systems that scale to hundreds or thousands of caches with thousands to millions of users. Rather than simply try to maximize hit rates, we take an end-to-end approach to improving response time by also considering hit times and miss times. We begin by studying several Internet caches and workloads, and we derive three core design principles for large scale distributed caches: (1) minimize the number of hops to locate and access data on both hits and misses, (2) share data among many users and scale to many caches, and (3) cache data close to clients. Our strategies for addressing these issues are built around a scalable, high-performance data-location service that tracks where objects are replicated. We describe how to construct such a service and how to use this service to provide direct access to remote data and push-based data replication. We evaluate our system through trace-driven simulation and find that these strategies together provide response time speedups of 1.27 to 2.43 compared to a traditional three-level cache hierarchy for a range of trace workloads and simulated environments. 1.

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.

This paper describes httperf, a tool for measuring web server performance. It provides a flexible facility for generating various HTTP workloads and for measuring server performance. The focus of httperf is not on implementing one particular benchmark but on providing a robust, high-performance tool that facilitates the construction of both micro- and macro-level benchmarks. The three distinguishing characteristics of httperf are its robustness, which includes the ability to generate and sustain server overload, support for the HTTP/1.1 protocol, and its extensibility to new workload generators and performance measurements. In addition to reporting on the design and implementation of httperf this paper also discusses some of the experiences and insights gained while realizing this tool. 1 Introduction A web system consists of a web server, a number of clients, and a network that connects the clients to the server. The protocol used to communicate between the client and server is HTTP [...

User-perceived latency is recognized as the central performance problem in the Web. We systematically measure factors contributing to this latency, across several locations. Our study reveals that DNS query times, TCP connection establishment, and start-of-session delays at HTTP servers, more so than transmission time, are major causes of long waits. Wait due to these factors also afflicts high-bandwidth users and has detrimental effect on perceived performance.

We present what we believe to be the first thorough characterization of live streaming media content delivered over the Internet. Our characterization of over five million requests spanning a 28-day period is done at three increasingly granular levels, corresponding to clients, sessions, and transfers. Our findings support two important conclusions. First, we show that the nature of interactions between users and objects is fundamentally different for live versus stored objects. Access to stored objects is user driven, whereas access to live objects is object driven. This reversal of active/passive roles of users and objects leads to interesting dualities. For instance, our analysis underscores a Zipf-like profile for user interest in a given object, which is to be contrasted to the classic Zipf-like popularity of objects for a given user. Also, our analysis reveals that transfer lengths are highly variable and that this variability is due to the stickiness of clients to a particular live object, as opposed to structural (size) properties of objects. Second, based on observations we make, we conjecture that the particular characteristics of live media access workloads are likely to be highly dependent on the nature of the live content being accessed. In our study, this dependence is clear from the strong temporal correlations we observed in the traces, which we attribute to the synchronizing impact of live content on access characteristics. Based on our analyses, we present a model for live media workload generation that incorporates many of our findings, and which we implement in GISMO [19].

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...