Request Inter-arrival Time Research Articles

Self-Similarity in the Representative File System Workloads: Analysis and Modeling

As one of the workload characteristics, the anomaly behaviors in real workload have been recognized as a critical requirement for file system design. In this paper, a set of traces collected from typically realistic file system have been analyzed. The correlation study of I/O request inter-arrival times shows that it is necessary to examine the self-similarity in file system workload. Then the phenomenon of self-similarity which had been initially observed in network and disk I/O workload has also been visually and statistically observed in file system workload. In addition, we implement an I/O series generator in which the inputs are the measured properties of the available trace data. Experimental results show that this model can accurately emulate the complex access arrival behaviors of real file systems, particularly the heavy-tail characteristics.

Modeling a non-stationary bots’ arrival process at an e-commerce Web site

The paper concerns the issue of modeling and generating a representative Web workload for Web server performance evaluation through simulation experiments. Web traffic analysis has been done from two decades, usually based on Web server log data. However, while the character of the overall Web traffic has been extensively studied and modeled, relatively few studies have been devoted to the analysis of Web traffic generated by Internet robots (Web bots). Moreover, the overwhelming majority of studies concern the traffic on non e-commerce websites. In this paper we address the problem of modeling a realistic arrival process of bots’ requests on an e-commerce Web server. Based on real log data for an online store, sessions generated by bots were reconstructed and their key features were analyzed, including the interarrival time of bot sessions, the number of HTTP requests per session, and the interarrival time of requests in session. To deal with the problem of non-stationarity of the Web traffic, chunks associated with times of day were distinguished based on the intensity of bot sessions’ arrivals and then features of sessions in individual time chunks were analyzed separately. Using regression analysis, a mathematical model of the bots’ traffic features was developed and implemented in a bot traffic generator. Our findings confirm the existence of a heavy-tail in bot traffic features’ distributions. The bots’ session interarrival times and request interarrival times are best modeled by a Weibull and a sigmoid distributions, respectively, while the model proposed for the numbers of requests per bot session is based on a hybrid function being a combination of one exponential and two normal distribution functions. The suitable fit of the model was confirmed by the high correlation of the real and model data. Furthermore, a visual inspection of the simulation results showed that the estimated values represent distributions close to those of the empirical data.

MITTS

Memory bandwidth severely limits the scalability and performance of multicore and manycore systems. Application performance can be very sensitive to both the delivered memory bandwidth and latency. In multicore systems, a memory channel is usually shared by multiple cores. Having the ability to precisely provision, schedule, and isolate memory bandwidth and latency on a per-core basis is particularly important when different memory guarantees are needed on a per-customer, per-application, or per-core basis. Infrastructure as a Service (IaaS) Cloud systems, and even general purpose multicores optimized for application throughput or fairness all benefit from the ability to control and schedule memory access on a fine-grain basis. In this paper, we propose MITTS (Memory Inter-arrival Time Traffic Shaping), a simple, distributed hardware mechanism which limits memory traffic at the source (Core or LLC). MITTS shapes memory traffic based on memory request inter-arrival time, enabling fine-grain bandwidth allocation. In an IaaS system, MITTS enables Cloud customers to express their memory distribution needs and pay commensurately. For instance, MITTS enables charging customers that have bursty memory traffic more than customers with uniform memory traffic for the same aggregate bandwidth. Beyond IaaS systems, MITTS can also be used to optimize for throughput or fairness in a general purpose multi-program workload. MITTS uses an online genetic algorithm to configure hardware bins, which can adapt for program phases and variable input sets. We have implemented MITTS in Verilog and have taped-out the design in a 25-core 32nm processor and find that MITTS requires less than 0.9% of core area. We evaluate across SPECint, PARSEC, Apache, and bhm Mail Server workloads, and find that MITTS achieves an average 1.18× performance gain compared to the best static bandwidth allocation, a 2.69× average performance/cost advantage in an IaaS setting, and up to 1.17× better throughput and 1.52× better fairness when compared to conventional memory bandwidth provisioning techniques.

DiskAccel

Disk traces are typically used to analyze real-life workloads and for replay-based evaluations. This approach benefits from capturing important details such as varying behavior patterns, bursty activity, and diurnal patterns of system activity, which are often missing from the behavior of workload synthesis tools. However, accurate capture of such details requires recording traces containing long durations of system activity, which are difficult to use for replay-based evaluation. One way of solving the problem of long storage trace duration is the use of disk simulators. While publicly available disk simulators can greatly accelerate experiments, they have not kept up with technological innovations in the field. The variety, complexity, and opaque nature of storage hardware make it very difficult to implement accurate simulators. The alternative, replaying the whole traces on real hardware, suffers from either long run-time or required manual reduction of experimental time, potentially at the cost of reduced accuracy. On the other hand, burstiness, auto-correlation, and complex spatio-temporal properties of storage workloads make the known methods of sampling workload traces less effective. In this paper, we present a methodology called DiskAccel to efficiently select key intervals of a trace as representatives and to replay them to estimate the response time of the whole workload. Our methodology extracts a variety of spatial and temporal features from each interval and uses efficient data mining techniques to select the representative intervals. To verify the proposed methodology, we have implemented a tool capable of running whole traces or selective intervals on real hardware, warming up hardware state in an accelerated manner, and emulating request causality while minimizing request inter-arrival time error. Based on our experiments, DiskAccel manages to speed up disk replay by more than two orders of magnitude, while keeping average estimation error at 7.6%.

A Distributed Probabilistic Commitment Control Algorithm for Service-Oriented Systems

Application creation through service composition is a cornerstone for several architectures including Service-Oriented Architecture. As the number and diversity of applications created based on this paradigm increase, the need for guaranteeing quality of service becomes more important. In this paper, we present a distributed algorithm for guaranteeing a specified level of application completion probability. The algorithm is designed to control service commitments in both queue-less and queue-enabled service-oriented systems. The algorithm does not assume a specific distribution type for service execution times and application request inter-arrival times, and hence is suitable for systems with stationary or non-stationary request arrivals. We show that the proposed distributed algorithm achieves its performance objectives for both queue-less and queue-enabled service oriented systems.

Dynamic traffic controls for Web-server networks

Distributed Web-server systems have been widely used to provide effective Internet services. The management of these systems requires dynamic controls of the Web traffic. With the development of multimedia Web sites and increasingly diversified services, the existing load balancing approaches can no longer satisfy the requirements of either the service providers or the users. In this paper, a new reward-based control mechanism is proposed that can satisfy the dynamic content-based control requirement while avoiding congestion at the dispatcher. The core of the control algorithm is based on an MDP model. To minimize the system overhead, a centralized dispatching with decentralized admission (CDDA) approach is used to distribute the control related computation to each server pool. This cuts down the dimensions of the problem dramatically. We also propose a state-block scheme to further reduce the state space so that the algorithm becomes computationally feasible for on-line implementation. Simulation results demonstrate that the proposed state-block approach can not only reduce the computation time dramatically but also provide a good approximation of power-tailed request interarrival times common for Internet traffic. Finally, an implementation plan with system design is also proposed.

상호참조시간을 고려한 단기간 임시지역성 측정

Temporal locality of Web server references is one of the important characteristics to be considered in the design of a Web caching strategy, and it is important to measure the temporal locality exactly. Various methods to estimate the temporal locality have been proposed, however, Web server designers have still troubles in its measurement by using the tools that don`t reflect the interarrival time of document requests. In this paper, we propose a measurement tool for short-term temporal locality based on request interarrival time, and discuss the simulation results based on the flares from NLANR and NASA Web sites. The results show that the proposed tool estimates the short-term temporal locality more exactly than that based on a stack.

Tune to Lambda patching

A recent paper by Hua, Cai and Sheu [7] describes Patching as a technique for reducing server load in a true video-on-demand (TVoD) system. It is a scheme for multicast video transmissions, which outperforms techniques such as Batching in response time and Piggybacking in bandwidth savings for titles of medium popularity, and probably in user satisfaction as well. It achieves TVoD performance by buffering part of the requested video in the receiving end-system.In a further study, the authors give analytical and simulation details on optimized patching windows under the assumptions of the Grace and Greedy patching techniques. In our view, this does not exploit fully the calculation that was performed in that study. We state that temporal distance between two multicast streams for one movie should not be determined by a client policy or simulation. Rather, it can be calculated by the server on a per video basis, since the server is aware of the average request interarrival time for each video. Since we model the request arrivals as a Poisson process, which is defined by a single variable that is historically called λ, we call this variation "λ Patching". Furthermore, we present an optimization option "Multistream Patching" that reduces the server load further. We accept that some near video-on-demand-like traffic is generated with additional patch streams, and achieve additional gains in server load.

Analysis and modeling of World Wide Web traffic for capacity dimensioning of Internet access lines

A study on traffic characterization of the Internet is essential to design the Internet infrastructure. In this paper, we first characterize WWW (World Wide Web) traffic based on the access log data obtained at four different servers. We find that the document size, the request inter-arrival time and the access frequency of WWW traffic follow heavy-tail distributions. Namely, the document size and the request inter-arrival time follow log-normal distributions, and the access frequency does the Pareto distribution. For the request inter-arrival time, however, an exponential distribution becomes adequate if we are concerned with the busiest hours. Based on our analytic results, we next build an M/G/1/PS queuing model to discuss a design methodology of the Internet access network. The accuracy of our model is validated by comparing with the trace-driven simulation. We also investigate the effect of document caching at the Proxy server on the WWW traffic characteristics. The results show that the traffic volume is actually reduced by the document replacement policies, but the traffic characteristics are not much affected. It suggests that our modeling approach can be applied to the case with document caching, which is demonstrated by simulation experiments.