ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)最新文献

{"title":"Investigating Characteristics of Internet Paths","authors":"Khalid Bakhshaliyev, M. A. Canbaz, M. H. Gunes","doi":"10.1145/3342286","DOIUrl":"https://doi.org/10.1145/3342286","url":null,"abstract":"Interactive and multimedia applications depend on the stability of end-to-end paths for predictable performance and good quality of service. On the other hand, network providers depend on multiple paths to ensure fault tolerance and use load balancing between these paths to enhance the overall network throughput. In this study, we analyze path dynamics for both end-to-end paths and path segments within service providers’ networks using 2 months of measurement data from the RIPE Atlas platform, which collects path traces between a fixed set of source and destination pairs every 15 minutes. We observe that 78% of the end-to-end routes have at least two alternative Autonomous System (AS) paths with some end-to-end routes going through hundreds of different AS paths during the 2 months of analysis. While AS level paths are often prevalent for a day, there are considerable changes in the routing of the trace packets over the ASes for a longer duration of a month or longer. Analyzing end-to-end paths for routing anomalies, we observe that 4.4% of the path traces (involving 18% of the ASes) contain routing loops indicating misconfiguration of routers. Some of the ASes had over 100 routers involved in loops in path traces through their networks. We observe a much higher rate of anomalies in the AS level, with 45% of path traces containing an AS loop. Additionally, we discovered that few of the ASes bounce-back packets where some traces through their network traverse routers in both forward and backward directions. Determining path segments belonging to each AS, we further explore ingress to egress paths of ASes in addition to the source to destination paths within the AS. Analyzing trace segments between ingresses and egresses of an AS, we realized more than half of the ASes have the same router level path between any ingress-egress pair for the 2 months, but others implement load balancing. These results are different from earlier studies that indicated a high level of path dynamism. Our results indicate that the end-to-end path dynamism is due to the Border Gateway Protocol level rather than the router level within ASes.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133840649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On the Endurance of the d-Choices Garbage Collection Algorithm for Flash-Based SSDs","authors":"Robin Verschoren, B. V. Houdt","doi":"10.1145/3326121","DOIUrl":"https://doi.org/10.1145/3326121","url":null,"abstract":"Garbage collection (GC) algorithms for flash-based solid-state drives (SSDs) have a profound impact on its performance and many studies have focused on assessing the so-called write amplification of various GC algorithms. In this article, we consider the family of d-choices GC algorithms and study (a) the extent in which these algorithms induce unequal wear and (b) the manner in which they affect the lifetime of the drive. For this purpose, we introduce two performance measures: PE fairness and SSD endurance. We study the impact of the d-choices GC algorithm on both these measures under different workloads (uniform, synthetic and trace-based) when combined with two different write modes. Numerical results show that the more complex of the two write modes, which requires hot/cold data identification, may not necessarily give rise to a significantly better SSD endurance. Further, the d-choices GC algorithm is often shown to strike a good balance between garbage collection and wear leveling for small d values (e.g., d = 10), yielding high endurance.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130199471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing N-Tier Application Scalability in the Cloud","authors":"Qingyang Wang, Shungeng Zhang, Yasuhiko Kanemasa, C. Pu, Balaji Palanisamy, L. Harada, Motoyuki Kawaba","doi":"10.1145/3326120","DOIUrl":"https://doi.org/10.1145/3326120","url":null,"abstract":"An effective cloud computing environment requires both good performance and high efficiency of computing resources. Through extensive experiments using a representative n-tier benchmark application (Rice University Bulletin Board System (RUBBoS)), we show that the soft resource allocation (e.g., thread pool size and database connection pool size) in component servers has a significant impact on the overall system performance, especially at high system utilization scenarios. Concretely, the same software resource allocation can be a good setting in one hardware configuration and then becomes an either under- or over-allocation in a slightly different hardware configuration, causing a significant performance drop. We have also observed some interesting phenomena that were caused by the non-trivial dependencies between the soft resources of servers in different tiers. For instance, the thread pool size in an Apache web server can limit the total number of concurrent requests to the downstream servers, which surprisingly decreases the Central Processing Unit (CPU) utilization of the Clustered Java Database Connectivity (C-JDBC) clustering middleware as the workload increases. To provide a globally optimal (or near-optimal) soft resource allocation of each tier in the system, we propose a practical iterative solution approach by combining a soft resource aware queuing network model and the fine-grained measurement data of every component server. Our results show that to truly scale complex distributed systems such as n-tier web applications with expected performance in the cloud, we need to carefully manage soft resource allocation in the system.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128433539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Performance of Redundancy(d) with Identical/Independent Replicas","authors":"T. Hellemans, B. V. Houdt","doi":"10.1145/3316768","DOIUrl":"https://doi.org/10.1145/3316768","url":null,"abstract":"Queueing systems with redundancy have received considerable attention recently. The idea of redundancy is to reduce latency by replicating each incoming job a number of times and to assign these replicas to a set of randomly selected servers. As soon as one replica completes service the remaining replicas are cancelled. Most prior work on queueing systems with redundancy assumes that the job durations of the different replicas are independent and identically distributed (i.i.d.), which yields insights that can be misleading for computer system design. In this article, we develop a differential equation, using the cavity method, to assess the workload and response time distribution in a large homogeneous system with redundancy without the need to rely on this independence assumption. More specifically, we assume that the duration of each replica of a single job is identical across the servers and follows a general service time distribution. Simulation results suggest that the differential equation yields exact results as the system size tends to infinity and can be used to study the stability of the system. We also compare our system to the one with i.i.d. replicas and show the similarity in the analysis used for independent, respectively, identical replicas.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115877962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A New Framework for Evaluating Straggler Detection Mechanisms in MapReduce","authors":"Tien D. Phan, Guillaume Pallez, Shadi Ibrahim, P. Raghavan","doi":"10.1145/3328740","DOIUrl":"https://doi.org/10.1145/3328740","url":null,"abstract":"Big Data systems (e.g., Google MapReduce, Apache Hadoop, Apache Spark) rely increasingly on speculative execution to mask slow tasks, also known as stragglers, because a job’s execution time is dominated by the slowest task instance. Big Data systems typically identify stragglers and speculatively run copies of those tasks with the expectation that a copy may complete faster to shorten job execution times. There is a rich body of recent results on straggler mitigation in MapReduce. However, the majority of these do not consider the problem of accurately detecting stragglers. Instead, they adopt a particular straggler detection approach and then study its effectiveness in terms of performance, e.g., reduction in job completion time or higher efficiency, e.g., high resource utilization. In this article, we consider a complete framework for straggler detection and mitigation. We start with a set of metrics that can be used to characterize and detect stragglers including Precision, Recall, Detection Latency, Undetected Time, and Fake Positive. We then develop an architectural model by which these metrics can be linked to measures of performance including execution time and system energy overheads. We further conduct a series of experiments to demonstrate which metrics and approaches are more effective in detecting stragglers and are also predictive of effectiveness in terms of performance and energy efficiencies. For example, our results indicate that the default Hadoop straggler detector could be made more effective. In a certain case, Precision is low and only 55% of those detected are actual stragglers and the Recall, i.e., percent of actual detected stragglers, is also relatively low at 56%. For the same case, the hierarchical approach (i.e., a green-driven detector based on the default one) achieves a Precision of 99% and a Recall of 29%. This increase in Precision can be translated to achieve lower execution time and energy consumption, and thus higher performance and energy efficiency; compared to the default Hadoop mechanism, the energy consumption is reduced by almost 31%. These results demonstrate how our framework can offer useful insights and be applied in practical settings to characterize and design new straggler detection mechanisms for MapReduce systems.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126463101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Production Application Performance Data Streaming for System Monitoring","authors":"Ramin Izadpanah, B. Allan, D. Dechev, J. Brandt","doi":"10.1145/3319498","DOIUrl":"https://doi.org/10.1145/3319498","url":null,"abstract":"In this article, we present an approach to streaming collection of application performance data. Practical application performance tuning and troubleshooting in production high-performance computing (HPC) environments requires an understanding of how applications interact with the platform, including (but not limited to) parallel programming libraries such as Message Passing Interface (MPI). Several profiling and tracing tools exist that collect heavy runtime data traces either in memory (released only at application exit) or on a file system (imposing an I/O load that may interfere with the performance being measured). Although these approaches are beneficial in development stages and post-run analysis, a systemwide and low-overhead method is required to monitor deployed applications continuously. This method must be able to collect information at both the application and system levels to yield a complete performance picture. In our approach, an application profiler collects application event counters. A sampler uses an efficient inter-process communication method to periodically extract the application counters and stream them into an infrastructure for performance data collection. We implement a tool-set based on our approach and integrate it with the Lightweight Distributed Metric Service (LDMS) system, a monitoring system used on large-scale computational platforms. LDMS provides the infrastructure to create and gather streams of performance data in a low overhead manner. We demonstrate our approach using applications implemented with MPI, as it is one of the most common standards for the development of large-scale scientific applications. We utilize our tool-set to study the impact of our approach on an open source HPC application, Nalu. Our tool-set enables us to efficiently identify patterns in the behavior of the application without source-level knowledge. We leverage LDMS to collect system-level performance data and explore the correlation between the system and application events. Also, we demonstrate how our tool-set can help detect anomalies with a low latency. We run tests on two different architectures: a system enabled with Intel Xeon Phi and another system equipped with Intel Xeon processor. Our overhead study shows our method imposes at most 0.5% CPU usage overhead on the application in realistic deployment scenarios.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133432124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Managing Response Time Tails by Sharding","authors":"P. Harrison, N. Patel, Juan F. Pérez, Z. Qiu","doi":"10.1145/3300143","DOIUrl":"https://doi.org/10.1145/3300143","url":null,"abstract":"Matrix analytic methods are developed to compute the probability distribution of response times (i.e., data access times) in distributed storage systems protected by erasure coding, which is implemented by sharding a data object into N fragments, only K<; N of which are required to reconstruct the object. This leads to a partial-fork-join model with a choice of canceling policies for the redundant N−K tasks. The accuracy of the analytical model is supported by tests against simulation in a broad range of setups. At increasing workload intensities, numerical results show the extent to which increasing the redundancy level reduces the mean response time of storage reads and significantly flattens the tail of their distribution; this is demonstrated at medium-high quantiles, up to the 99th. The quantitative reduction in response time achieved by two policies for canceling redundant tasks is also shown: for cancel-at-finish and cancel-at-start, which limits the additional load introduced whilst losing the benefit of selectivity amongst fragment service times.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123349118","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Provisioning and Performance Evaluation of Parallel Systems with Output Synchronization","authors":"Wasiur R. KhudaBukhsh, Sounak Kar, Amr Rizk, H. Koeppl","doi":"10.1145/3300142","DOIUrl":"https://doi.org/10.1145/3300142","url":null,"abstract":"Parallel server frameworks are widely deployed in modern large-data processing applications. Intuitively, splitting and parallel processing of the workload provides accelerated application response times and scaling flexibility. Examples of such frameworks include MapReduce, Hadoop, and Spark. For many applications, the dynamics of such systems are naturally captured by a Fork-Join (FJ) queuing model, where incoming jobs are split into tasks each of which is mapped to exactly one server. When all the tasks that belong to one job are executed, the job is reassembled and leaves the system. We consider this behavior at the output as a synchronization constraint. In this article, we study the performance of such parallel systems for different server properties, e.g., work-conservingness, phase-type behavior, and as suggested by recent evidence, for bursty input job arrivals. We establish a Large Deviations Principle for the steady-state job waiting times in an FJ system based on Markov-additive processes. Building on that, we present a performance analysis framework for FJ systems and provide computable bounds on the tail probabilities of the steady-state waiting times. We validate our bounds using estimates obtained through simulations. In addition, we define and analyze provisioning, a flexible division of jobs into tasks, in FJ systems. Finally, we use this framework together with real-world traces to show the benefits of an adaptive provisioning system that adjusts the service within an FJ system based on the arrival intensity.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128410330","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Framework for Allocating Server Time to Spot and On-Demand Services in Cloud Computing","authors":"Xiaohu Wu, F. Pellegrini, Guanyu Gao, G. Casale","doi":"10.1145/3366682","DOIUrl":"https://doi.org/10.1145/3366682","url":null,"abstract":"Cloud computing delivers value to users by facilitating their access to servers at any time period needed. An approach is to provide both on-demand and spot services on shared servers. The former allows users to access servers on demand at a fixed price and users occupy different time periods on servers. The latter allows users to bid for the remaining unoccupied time periods via dynamic pricing; however, without appropriate design, such time periods may be arbitrarily short since on-demand users arrive randomly. This is also the current service model adopted by Amazon Elastic Cloud Compute. In this article, we provide the first integral framework for sharing time on servers between on-demand and spot services while optimally pricing spot service. It guarantees that on-demand users can get served quickly while spot users can stably use servers for a properly long period once accepted, which is a key feature in making both on-demand and spot services accessible. Simulation results show that, by complementing the on-demand market with a spot market, a cloud provider can improve revenue by up to 461.5%. The framework is designed under assumptions that are met in real environments. It is a new tool that other cloud operators can use to quantify the advantage of a hybrid spot and on-demand service, making the case for eventually integrating this service model into their own infrastructures.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"83 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132107723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"EnergyQARE","authors":"Hao Chen, Yijia Zhang, M. Caramanis, A. Coskun","doi":"10.1145/3243172","DOIUrl":"https://doi.org/10.1145/3243172","url":null,"abstract":"Power market operators have recently introduced smart grid demand response (DR), in which electricity consumers regulate their power usage following market requirements. DR helps stabilize the grid and enables integrating a larger amount of intermittent renewable power generation. Data centers provide unique opportunities for DR participation due to their flexibility in both workload servicing and power consumption. While prior studies have focused on data center participation in legacy DR programs such as dynamic energy pricing and peak shaving, this article studies data centers in emerging DR programs, i.e., demand side capacity reserves. Among different types of capacity reserves, regulation service reserves (RSRs) are especially attractive due to their relatively higher value. This article proposes EnergyQARE, the Energy and Quality-of-Service (QoS) Aware RSR Enabler, an approach that enables data center RSR provision in real-life scenarios. EnergyQARE not only provides a bidding strategy in RSR provision, but also contains a runtime policy that adaptively modulates data center power through server power management and server provisioning based on workload QoS feedback. To reflect real-life scenarios, this runtime policy handles a heterogeneous set of jobs and considers transition time delay of servers. Simulated numerical results demonstrate that in a general data center scenario, EnergyQARE provides close to 50% of data center average power consumption as reserves to the market and saves up to 44% in data center electricity cost, while still meeting workload QoS constraints. Case studies in this article show that the percentages of savings are not sensitive to a specific type of non-interactive workload, or the size of the data center, although they depend strongly on data center utilization and parameters of server power states.","PeriodicalId":105474,"journal":{"name":"ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114520372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}