We developed a SPARC-V9 processor, the SPARC64 V. It has an operating frequency of 1.35 GHz and contains 191 million transistors fabricated using 0.13-µm CMOS technology with eight-layer copper metallization. SPECjbb2000 (CPU# 32) is 492683, highest on the market and 42% higher than the next highest system. SPEC CPU2000 performance is 858 for SPECint and 1228 for SPECfp. The processor is designed to provide the high system performance and high reliability required of enterprise server systems. It is also designed to address the performance requirements of high-performance computing. During our development of several generations of mainframe processors, we conducted many related experiments, and obtained enterprise server system (EPS) development skills, an understanding of EPS workload characteristics, and technology that provides high reliability, availability, and serviceability. We used those as bases of the new processor development. The approach quite effectively moves beyond differences between mainframe and SPARC systems. At the beginning of development and before the start of hardware design, we developed a software performance simulator so we could understand the performance impacts of created specifications, thereby enabling us to make appropriate decisions about hardware design. We took this approach to solve performance problems before tape-out and avoid spending additional time on design update and physical machine reconstruction. We were successful, completing the high-performance processor development on schedule and in a short time. This paper describes the SPARC64 V microprocessor and performance analyses for development of its design.

As so many software titles are now being distributed via the Internet, the number of accesses to file servers, such as FTP servers, is rapidly increasing. To prevent the concentration of accesses to the original file server, mirror servers are being introduced that contain the same directories and files as held by the original server. However, inconsistency among the mirror servers and the original server is often observed because of delivery latency, traffic congestion on the network, and management policies of the mirror servers. This inconsistency degrades the value of the mirror servers. Accordingly, we have developed an intermediate FTP proxy server system that guarantees the freshness of the files as well as preventing access concentration on the original FTP server. The system adopts per-file selection of the replicated files; most existing methods are based on per-host or per-directory selection. Therefore it can assure users of a quick, stable, and up-to-date FTP mirroring service even in the face of frequent content updates, which tend to degrade the homogeneity of services. Moreover, it can forward the retrieved files with little overhead. Tests confirmed that our system is comparable to existing systems from the viewpoint of actual retrieval time, required traffic, and load endurance. This technology can assure clients that they will receive the latest version of the file(s) desired. It well supports heterogeneous network environments such as the Internet.

In this letter, a new output feedback tracking control using a fuzzy disturbance observer (FDO) is proposed and its application to control of a nonlinear system in the presence of the internal parameter perturbation and external disturbance is presented. An FDO using a filtered signal is developed and the high gain observer (HGO) is employed to implement the output feedback tracking control. It is shown in a rigorous manner that all the errors involved can be kept arbitrarily small. Finally, the effectiveness and the feasibility of the suggested method is demonstrated by computer simulation.

As a lot of programs and contents such as movie files are being delivered via the Internet, and copies are often stored in distributed servers in order to reduce the load on the original servers, to ease network congestion, and to decrease response time. To retrieve an object file, existing methods simply select one or more servers. Such methods divide a file into equal pieces whose size is determined a priori. This approach is not practical for networks that offer variable bandwidth. In order to more utilize variable bandwidth, we propose an adaptive downloading method. We evaluate it by experiments conducted on the Internet. The results show that the new method is effective and that it will become an important network control technology for assurance.

In this paper, we present an output feedback controller using a fuzzy controller and observer for nonlinear systems with unknown time-delay. Recently, Cao et al. proposed a stabilization method for the nonlinear time-delay systems using a fuzzy controller when the time-delay is known. In general, however, it is impossible to know or measure this time-varying delay. The proposed method requires only the upper bound of the derivative of the time-delay. We represent the nonlinear system with the unknown time-delay by Takagi-Sugeno (T-S) fuzzy model and design the fuzzy controller and observer for the systems using the parallel distributed compensation (PDC) scheme. In addition, we derive the sufficient condition for the asymptotic stability of the equilibrium point by applying Lyapunov-Krasovskii theorem to the closed-loop system and solve the condition in the formulation of LMI. Finally, computer simulations are included to demonstrate the effectiveness of the suggested method.

This paper considers a transaction-based multi-server system with rejuvenation, and derive the optimal software rejuvenation policies under some system dependability measures; the steady-state availability, the probability of transaction loss and the upper bound of mean response time on transactions. We compare the system configuration based on a single-server with that based on a multi-server in terms of the software rejuvenation scheme. In numerical examples, we calculate the optimal software rejuvenation timing and its associated dependability measure, and refer to the effect of preventive maintenance in the transaction-based multi-server software systems.

In content distribution networks (CDNs), the content routing which directs user requests to an adequate server from the viewpoint of improvement of latency for obtaining contents is one of the most important technical issues. Several information, e.g. server load or network delay, can be used for content routing. Network support, e.g. active network, enables a router to select an adequate server by using these information. In the paper, we investigate a server selection policy of a network support approach from the viewpoint of which information to be used for effective server selection. We propose a server selection policy using RTT information measured at a router. Simulation results show that our proposed server selection policy in content routing selects a good server under both conditions where server latency and network delay is a dominant element of user response time. Furthermore, we also investigate about location of routers with network support bringing good performance for our proposed scheme.

Load balancing among multiple mirror servers located at distributed positions in the network is a key technique for content delivery services. For bandwidth allocated services, we consider how to select a suitable server from several candidates containing the same content at the time of a request. We propose limiting the candidates in advance and selecting a server from the limited set of servers in a round-robin fashion. The server sets that minimize the variance of the link load are derived using a greedy method for a given network topology and service demand. Through numerical evaluation, we demonstrate that the proposed method is superior to previous methods.

In this paper, we present an Internet personal live-broadcasting server system. Our solution is not only for experts but also for amateur users who want to broadcast using simple multimedia equipment. For scalable broadcasting services, we developed multiple-channel establishment and channel expansion. Concurrent services for a large number of broadcasting channels are effectively provided. Also, the capacity of channels can be expanded as the number of participants increases. Furthermore, for the sake of complete live broadcasting with high-quality transmission, the system supports both TCP (transmission control protocol) and UDP (user datagram protocol) according to the status of network environments as well as the received packet loss in the user system. The performance of the system is effectively evaluated at such practical commercial sites as well-known community and E-business sites.

This paper presents an observer design methodology for a special class of MIMO nonlinear systems. First, we characterize the class of MIMO nonlinear systems that consists of the linear observable part and the nonlinear part with a block triangular structure. Also, the similarity transformation that plays an important role in proving the convergence of the proposed observer is generalized to MIMO systems. Since the gain of the proposed observer minimizes a nonlinear part of the system to suppress for the stability of the error dynamics, it improves the transient performance of the high gain observer. Moreover, by using the generalized similarity transformation, it is shown that under some observability and boundedness conditions, the proposed observer guarantees the global exponential convergence to zero of the estimation error. Finally, the simulation results for induction motor are included to illustrate the validity of our design scheme.

This letter presents the implementation framework of a video streaming server which uses an optical disk library as a source of media archiving. In order to handle the optical storage subsystem in the framework of disk-based stream service model, we have devised an effective stream scheduling, disk caching, and admission control mechanism. The proposed system has been implemented and its key principles are validated with real experiments.

Many services on the Internet are provided by multiple identical servers in order to improve performance and robustness. The number, the location and the distribution of servers affect the performance and reliability of a service. The server placement is, however, often determined based on the empirical knowledge of the administrators. This paper investigates issues of the server placement in terms of the service performance and the server load. We identify that a server selection mechanism plays an important role in server placement, and thus, evaluate different server selection algorithms. The result shows that it is essential to the robustness of a service to employ a mechanism which distributes service requests to the servers according to the measured response time of each server. As a case study, we evaluate the server selection mechanisms employed by different DNS (Domain Name System) implementations. Then, we show the effects of the different server selection algorithms using root-server measurements taken at different locations around the world.

In this paper, we propose a new job scheduling method for distributed parallel systems that can simultaneously achieve two main goals of the job scheduling in those systems: to minimize the execution time of a parallel job without disturbing the execution of the other jobs. We try to achieve those goals by introducing a new scheduler, called active scheduler, that dynamically controls the priority of parallel programs and balances the workload of host computers depending on the status of the underlying runtime environment. We implemented a prototype system of the scheduler to evaluate its effectiveness. The result of experiments implies that the overhead of introducing the active scheduler is at most 15% of the original execution time, and it is in fact effective to adjust the execution of parallel programs to an actual distributed environment in which many users execute their jobs simultaneously.

Intensive accesses to a web server causes its response delay and/or service suspension. We propose, to solve the problem in this paper, a service model for web servers grounded upon admission control and call waiting service, which are common approaches to designing telephone systems. Software architecture which consists of a front-end server and child servers is also designed based on the service model. Through performance evaluation for an implementation of the architecture, it is shown that the overall throughput becomes much better by keeping child web servers from overload. We drew a conclusion that the architecture enhances flexibility of a resulting software with discussions from the extensibility view.

Server performance is a major issue in improving the overall performance of the World Wide Web (WWW). This article introduces a dynamic mirroring-based approach to improve WWW servers' performance. In contrast to static mirroring, where mirror servers are allocated statically, our mirror servers' setup is driven by network traffic measurement. Performance in terms of latency is inferred from a queuing model. According to this model we show that latency of an overloaded server can be tuned by delegating a portion of the load to a cooperative mirror server. Cost is evaluated by the amount of load hosted by the mirror servers. The goal is then to keep the latency within a tolerable threshold, while minimizing the delegated load. This problem is formulated as a constrained optimization problem where the task is to assign a portion of load corresponding to each document to each mirror server. As the result of this work, we will have a balanced load among the servers, and a smoother traffic along the Internet, as well. Empirical results show that this approach can guarantee to maintain the performance while showing a significant decrease in the amount of load transferred to the mirror servers.

The explosive growth of World Wide Web usage is causing a number of performance problems, including slow response times, network congestion, and denial of service. Web site that has a huge number of accesses and requires high quality of services, such as a site offering hosting services, or content delivery services, usually uses a cache server to reduce the load on the original server offering the original content. To increase the throughput of the caching process and to improve service availability, multiple cache servers are often positioned in front of the original server. This requires a switch to direct incoming requests to one of the multiple cache servers. In this paper, we propose a routing algorithm for such a switch in front of clustered multiple cache servers and evaluate its performance by simulation. The results show that our routing algorithm is effective when content has request locality and a short period of validity, for example, news, map data, road traffic data, or weather information. We also identify points to consider when the proposed algorithm is applied to a real system.

This paper describes the adaptability of communication software through a biologically-inspired policy coordination. Many research efforts have developed adaptable systems that allow various users or applications to meet their specific requirements by configuring different design and optimization policies. Navigating through many policies manually, however, is tedious and error-prone. Developers face the significant manual and ad-hoc work of engineering an system. In contrast, we propose to provide autonomous adaptability in communication endsystem with OpenWebServer/iNexus, which is both a web server and an object-oriented framework to tailer various web services and applications. The OpenWebServer's modular architecture allows to abstract and maintain a wide range of aspects in a HTTP server, and reconfigure the system by adding, deleting, changing, or replacing their policies. iNexus is a tool for automated policy-based management of OpenWebServer. Its design is inspired by the natural immune system, particularly immune network, a truly autonomous decentralized system. iNexus inspects the current system condition of OpenWebServer periodically, measures the delivered quality of service, and selects suitable set of policies to reconfigure the system dynamically by relaxing constraints between them. The policy coordination process is performed through decentralized interactions among policies without a single point of control, as the natural immune system does. This paper discusses communication software can evolve continuously in the piecemeal way with biological concepts and mechanisms, adapting itself to ever-changing environment.

The authors have been in charge of the operation of one of the root DNS servers for more than three years. In this paper, the overview of our system to provide high availability is introduced. In the following sections, a traffic analysis system to analyze the characteristics of the DNS queries and the brief summary which may help future DNS system deployment is described.

In server load balancing where replicated servers are dispersed geographically and accesses from clients are distributed to replicated servers, a way of distributing the accesses from clients to an adequate server plays an important role from the viewpoint of load balancing. In the paper, we propose a new network paradigm for server load balancing using active anycast. In active anycast, an end user only sends its request to group of servers using an anycast address. When this request arrives at an active router, it selects an adequate server from the viewpoint of load balancing and changes the anycast address of a packet to the unicast address of the selected server. The decision which server is the best one from the viewpoint of server load balancing is made by an active router rather than an end user, so active anycast is a network-initiated method. Simulation results show that active anycast can accomplish efficient server load balancing, even when a small part of routers are equipped with active network technology.

In this paper, we proposed two models, the full multiple MCS (Multicast Server) model and the hybrid multiple MCS model to support multiple MCS over a single large cluster in ATM (Asynchronous Transfer Mode) networks. Also, we presented two methods for MCS assignment which are known as 2PSPMT (2 Phase Shortest Path based on Multicast tree) and hybrid-2PSPMT, and evaluated its performance by simulation. When an ATM host requests joining a specific multicast group, the MARS (Multicast Address Resolution Server) designates a proper MCS among the multiple MCSs for the group member to minimize the average path delay between the sender and the group members. Each method for MCS assignment construct a 2-phase partial multicast tree based on the shortest path algorithm. We reduced the average path delay in the multicast tree using these methods with various cluster topologies and MCS distribution scenarios in addition to distributing the load among multiple MCSs.