Unikernels are a promising alternative for application deployment in cloud platforms. They comprise a very small footprint, providing better deployment agility and portability among virtualization platforms. Similar to Linux containers, they are a lightweight alternative for deploying distributed applications based on microservices. However, the comparison of unikernels with other virtualization options regarding the concurrent provisioning of instances, as in the case of microservices-based applications, is still lacking. This paper provides an evaluation of KVM (Virtual Machines), Docker (Containers), and OSv (Unikernel), when provisioning multiple instances concurrently in an OpenStack cloud platform. We confirmed that OSv outperforms the other options and also identified opportunities for optimization.
{"title":"Time Provisioning Evaluation of KVM, Docker and Unikernels in a Cloud Platform","authors":"Bruno Xavier, T. Ferreto, L. C. Jersak","doi":"10.1109/CCGrid.2016.86","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.86","url":null,"abstract":"Unikernels are a promising alternative for application deployment in cloud platforms. They comprise a very small footprint, providing better deployment agility and portability among virtualization platforms. Similar to Linux containers, they are a lightweight alternative for deploying distributed applications based on microservices. However, the comparison of unikernels with other virtualization options regarding the concurrent provisioning of instances, as in the case of microservices-based applications, is still lacking. This paper provides an evaluation of KVM (Virtual Machines), Docker (Containers), and OSv (Unikernel), when provisioning multiple instances concurrently in an OpenStack cloud platform. We confirmed that OSv outperforms the other options and also identified opportunities for optimization.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"15 11","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120921743","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}
Xiangyu Guo, Binbin Tang, Jian Tao, Zhaohui Huang, Zhihui Du
PPMLR-MHD is a new magnetohydrodynamics (MHD) model used to simulate the interactions of the solar wind with the magnetosphere, which has been proved to be the key element of the space weather cause-and-effect chain process from the Sun to Earth. Compared to existing MHD methods, PPMLR-MHD achieves the advantage of high order spatial accuracy and low numerical dissipation. However, the accuracy comes at a cost. On one hand, this method requires more intensive computation. On the other hand, more boundary data is subject to be transferred during the process of simulation. In this work, we present a parallel hybrid solution of the PPMLR-MHD model implemented using the computing capabilities of both CPUs and GPUs. We demonstrate that our optimized implementation alleviates the data transfer overhead by using GPU Direct technology and can scale up to 151 processes and achieve significant performance gains by distributing the workload among the CPUs and GPUs on Titan at Oak Ridge National Laboratory. The performance results show that our implementation is fast enough to carry out highly accurate MHD simulations in real time.
{"title":"Large Scale GPU Accelerated PPMLR-MHD Simulations for Space Weather Forecast","authors":"Xiangyu Guo, Binbin Tang, Jian Tao, Zhaohui Huang, Zhihui Du","doi":"10.1109/CCGrid.2016.68","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.68","url":null,"abstract":"PPMLR-MHD is a new magnetohydrodynamics (MHD) model used to simulate the interactions of the solar wind with the magnetosphere, which has been proved to be the key element of the space weather cause-and-effect chain process from the Sun to Earth. Compared to existing MHD methods, PPMLR-MHD achieves the advantage of high order spatial accuracy and low numerical dissipation. However, the accuracy comes at a cost. On one hand, this method requires more intensive computation. On the other hand, more boundary data is subject to be transferred during the process of simulation. In this work, we present a parallel hybrid solution of the PPMLR-MHD model implemented using the computing capabilities of both CPUs and GPUs. We demonstrate that our optimized implementation alleviates the data transfer overhead by using GPU Direct technology and can scale up to 151 processes and achieve significant performance gains by distributing the workload among the CPUs and GPUs on Titan at Oak Ridge National Laboratory. The performance results show that our implementation is fast enough to carry out highly accurate MHD simulations in real time.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116704950","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}
J. Kwak, Eunji Hwang, Tae-kyung Yoo, Beomseok Nam, Young-ri Choi
In this paper, we investigate techniques to effectively orchestrate HDFS in-memory caching for Hadoop. We first evaluate a degree of benefit which each of various MapReduce applications can get from in-memory caching, i.e. cache affinity. We then propose an adaptive cache local scheduling algorithm that adaptively adjusts the waiting time of a MapReduce job in a queue for a cache local node. We set the waiting time to be proportional to the percentage of cached input data for the job. We also develop a cache affinity cache replacement algorithm that determines which block is cached and evicted based on the cache affinity of applications. Using various workloads consisting of multiple MapReduce applications, we conduct experimental study to demonstrate the effects of the proposed in-memory orchestration techniques. Our experimental results show that our enhanced Hadoop in-memory caching scheme improves the performance of the MapReduce workloads up to 18% and 10% against Hadoop that disables and enables HDFS in-memory caching, respectively.
{"title":"In-Memory Caching Orchestration for Hadoop","authors":"J. Kwak, Eunji Hwang, Tae-kyung Yoo, Beomseok Nam, Young-ri Choi","doi":"10.1109/CCGrid.2016.73","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.73","url":null,"abstract":"In this paper, we investigate techniques to effectively orchestrate HDFS in-memory caching for Hadoop. We first evaluate a degree of benefit which each of various MapReduce applications can get from in-memory caching, i.e. cache affinity. We then propose an adaptive cache local scheduling algorithm that adaptively adjusts the waiting time of a MapReduce job in a queue for a cache local node. We set the waiting time to be proportional to the percentage of cached input data for the job. We also develop a cache affinity cache replacement algorithm that determines which block is cached and evicted based on the cache affinity of applications. Using various workloads consisting of multiple MapReduce applications, we conduct experimental study to demonstrate the effects of the proposed in-memory orchestration techniques. Our experimental results show that our enhanced Hadoop in-memory caching scheme improves the performance of the MapReduce workloads up to 18% and 10% against Hadoop that disables and enables HDFS in-memory caching, respectively.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116725936","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}
One of the most important aspects that influences the performance of parallel applications is the speed of communication between their tasks. To optimize communication, tasks that exchange lots of data should be mapped to processing units that have a high network performance. This technique is called communication-aware task mapping and requires detailed information about the underlying network topology for an accurate mapping. Previous work on task mapping focuses on network clusters or shared memory architectures, in which the topology can be determined directly from the hardware environment. Cloud computing adds significant challenges to task mapping, since information about network topologies is not available to end users. Furthermore, the communication performance might change due to external factors, such as different usage patterns of other users. In this paper, we present a novel solution to perform communication-aware task mapping in the context of commercial cloud environments with multiple instances. Our proposal consists of a short profiling phase to discover the network topology and speed between cloud instances. The profiling can be executed before each application start as it causes only a negligible overhead. This information is then used together with the communication pattern of the parallel application to group tasks based on the amount of communication and to map groups with a lot of communication between them to cloud instances with a high network performance. In this way, application performance is increased, and data traffic between instances is reduced. We evaluated our proposal in a public cloud with a variety of MPI-based parallel benchmarks from the HPC domain, as well as a large scientific application. In the experiments, we observed substantial performance improvements (up to 11 times faster) compared to the default scheduling policies.
{"title":"Automatic Communication Optimization of Parallel Applications in Public Clouds","authors":"E. Carreño, M. Diener, E. Cruz, P. Navaux","doi":"10.1109/CCGrid.2016.59","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.59","url":null,"abstract":"One of the most important aspects that influences the performance of parallel applications is the speed of communication between their tasks. To optimize communication, tasks that exchange lots of data should be mapped to processing units that have a high network performance. This technique is called communication-aware task mapping and requires detailed information about the underlying network topology for an accurate mapping. Previous work on task mapping focuses on network clusters or shared memory architectures, in which the topology can be determined directly from the hardware environment. Cloud computing adds significant challenges to task mapping, since information about network topologies is not available to end users. Furthermore, the communication performance might change due to external factors, such as different usage patterns of other users. In this paper, we present a novel solution to perform communication-aware task mapping in the context of commercial cloud environments with multiple instances. Our proposal consists of a short profiling phase to discover the network topology and speed between cloud instances. The profiling can be executed before each application start as it causes only a negligible overhead. This information is then used together with the communication pattern of the parallel application to group tasks based on the amount of communication and to map groups with a lot of communication between them to cloud instances with a high network performance. In this way, application performance is increased, and data traffic between instances is reduced. We evaluated our proposal in a public cloud with a variety of MPI-based parallel benchmarks from the HPC domain, as well as a large scientific application. In the experiments, we observed substantial performance improvements (up to 11 times faster) compared to the default scheduling policies.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"301 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116254966","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}
V. Hendrix, James Fox, D. Ghoshal, L. Ramakrishnan
The growth in scientific data volumes has resulted in the need for new tools that enable users to operate on and analyze data on large-scale resources. In the last decade, a number of scientific workflow tools have emerged. These tools often target distributed environments, and often need expert help to compose and execute the workflows. Data-intensive workflows are often ad-hoc, they involve an iterative development process that includes users composing and testing their workflows on desktops, and scaling up to larger systems. In this paper, we present the design and implementation of Tigres, a workflow library that supports the iterative workflow development cycle of data-intensive workflows. Tigres provides an application programming interface to a set of programming templates i.e., sequence, parallel, split, merge, that can be used to compose and execute computational and data pipelines. We discuss the results of our evaluation of scientific and synthetic workflows showing Tigres performs with minimal template overheads (mean of 13 seconds over all experiments). We also discuss various factors (e.g., I/O performance, execution mechansims) that affect the performance of scientific workflows on HPC systems.
{"title":"Tigres Workflow Library: Supporting Scientific Pipelines on HPC Systems","authors":"V. Hendrix, James Fox, D. Ghoshal, L. Ramakrishnan","doi":"10.1109/CCGrid.2016.54","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.54","url":null,"abstract":"The growth in scientific data volumes has resulted in the need for new tools that enable users to operate on and analyze data on large-scale resources. In the last decade, a number of scientific workflow tools have emerged. These tools often target distributed environments, and often need expert help to compose and execute the workflows. Data-intensive workflows are often ad-hoc, they involve an iterative development process that includes users composing and testing their workflows on desktops, and scaling up to larger systems. In this paper, we present the design and implementation of Tigres, a workflow library that supports the iterative workflow development cycle of data-intensive workflows. Tigres provides an application programming interface to a set of programming templates i.e., sequence, parallel, split, merge, that can be used to compose and execute computational and data pipelines. We discuss the results of our evaluation of scientific and synthetic workflows showing Tigres performs with minimal template overheads (mean of 13 seconds over all experiments). We also discuss various factors (e.g., I/O performance, execution mechansims) that affect the performance of scientific workflows on HPC systems.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126785124","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}
K. Krish, Bharti Wadhwa, M. S. Iqbal, M. Mustafa Rafique, Ali R. Butt
A promising trend in storage management for big data frameworks, such as Hadoop and Spark, is the emergence of heterogeneous and hybrid storage systems that employ different types of storage devices, e.g. SSDs, RAMDisks, etc., alongside traditional HDDs. However, scheduling data accesses or requests to an appropriate storage device is non-trivial and depends on several factors such as data locality, device performance, and application compute and storage resources utilization. To this end, we present DUX, an application-attuned dynamic data management system for data processing frameworks, which aims to improve overall application I/O throughput by efficiently using SSDs only for workloads that are expected to benefit from them rather than the extant approach of storing a fraction of the overall workloads in SSDs. The novelty of DUX lies in profiling application performance on SSDs and HDDs, analyzing the resulting I/O behavior, and considering the available SSDs at runtime to dynamically place data in an appropriate storage tier. Evaluation of DUX with trace-driven simulations using synthetic Facebook workloads shows that even when using 5.5× fewer SSDs compared to a SSD-only solution, DUX incurs only a small (5%) performance overhead, and thus offers an affordable and efficient storage tier management.
{"title":"On Efficient Hierarchical Storage for Big Data Processing","authors":"K. Krish, Bharti Wadhwa, M. S. Iqbal, M. Mustafa Rafique, Ali R. Butt","doi":"10.1109/CCGrid.2016.61","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.61","url":null,"abstract":"A promising trend in storage management for big data frameworks, such as Hadoop and Spark, is the emergence of heterogeneous and hybrid storage systems that employ different types of storage devices, e.g. SSDs, RAMDisks, etc., alongside traditional HDDs. However, scheduling data accesses or requests to an appropriate storage device is non-trivial and depends on several factors such as data locality, device performance, and application compute and storage resources utilization. To this end, we present DUX, an application-attuned dynamic data management system for data processing frameworks, which aims to improve overall application I/O throughput by efficiently using SSDs only for workloads that are expected to benefit from them rather than the extant approach of storing a fraction of the overall workloads in SSDs. The novelty of DUX lies in profiling application performance on SSDs and HDDs, analyzing the resulting I/O behavior, and considering the available SSDs at runtime to dynamically place data in an appropriate storage tier. Evaluation of DUX with trace-driven simulations using synthetic Facebook workloads shows that even when using 5.5× fewer SSDs compared to a SSD-only solution, DUX incurs only a small (5%) performance overhead, and thus offers an affordable and efficient storage tier management.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126474733","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}
As we are facing ever increasing air traffic demand, it is critical to enhance air traffic capacity and alleviate humancontrollers' workload by viewing air traffic optimization as acontinuous/online streaming problem. Air traffic optimizationis commonly formulated as an integer linear programming(ILP) problem. Since ILP is NP-hard, it is computationallyintractable. Moreover, a fluctuating number of flights changescomputational demand dynamically. In this paper, we presentan elastic middleware framework that is specifically designedto solve ILP problems generated from continuous air trafficstreams. Experiments show that our VM scheduling algorithmwith time-series prediction can achieve similar performanceto a static schedule while using 49% fewer VM hours for arealistic air traffic pattern.
{"title":"Elastic Virtual Machine Scheduling for Continuous Air Traffic Optimization","authors":"Shigeru Imai, S. Patterson, Carlos A. Varela","doi":"10.1109/CCGrid.2016.87","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.87","url":null,"abstract":"As we are facing ever increasing air traffic demand, it is critical to enhance air traffic capacity and alleviate humancontrollers' workload by viewing air traffic optimization as acontinuous/online streaming problem. Air traffic optimizationis commonly formulated as an integer linear programming(ILP) problem. Since ILP is NP-hard, it is computationallyintractable. Moreover, a fluctuating number of flights changescomputational demand dynamically. In this paper, we presentan elastic middleware framework that is specifically designedto solve ILP problems generated from continuous air trafficstreams. Experiments show that our VM scheduling algorithmwith time-series prediction can achieve similar performanceto a static schedule while using 49% fewer VM hours for arealistic air traffic pattern.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126549870","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}
Dongju Chae, Joonsung Kim, Youngsok Kim, Jangwoo Kim, Kyung-Ah Chang, Sang-Bum Suh, Hyogun Lee
Application caching is a key feature to enable fast application switches for mobile devices by caching the entire memory pages of applications in the device's physical memory. However, application caching requires a prohibitive amount of memory unless a swap feature is employed to maintain only the working sets of the applications in memory. Unfortunately, mobile devices often disable the invaluable swap feature as it can severely decrease the flash-based local storage device's already marginal lifespan due to the increased writes to the device. As a result, modern mobile devices suffering from the insufficient memory space end up killing memory-hungry applications and keeping only a few applications in the memory. In this paper, we propose CloudSwap, a fast and robust swap mechanism for mobile devices to enable the memory-oblivious application caching. The key idea of CloudSwap is to use the fast local storage as a cache of read-intensive swap pages, while storing prefetch-enabled, write-intensive swap pages in a cloud storage. To preserve the lifespan of the local storage, CloudSwap minimizes the number of writes to the local storage by storing the modified portions of the locally swapped pages in a cloud. To reduce the remote swap-in latency, CloudSwap exploits two cloud-assisted prefetch schemes, the app-aware read-ahead scheme and the access pattern-aware prefetch scheme. Our evaluation shows that the performance of CloudSwap is comparable to a fast, but lifespan-critical local swap system, with only 18% lifespan reduction, compared to the local swap system's 85% lifespan reduction.
{"title":"CloudSwap: A Cloud-Assisted Swap Mechanism for Mobile Devices","authors":"Dongju Chae, Joonsung Kim, Youngsok Kim, Jangwoo Kim, Kyung-Ah Chang, Sang-Bum Suh, Hyogun Lee","doi":"10.1109/CCGrid.2016.22","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.22","url":null,"abstract":"Application caching is a key feature to enable fast application switches for mobile devices by caching the entire memory pages of applications in the device's physical memory. However, application caching requires a prohibitive amount of memory unless a swap feature is employed to maintain only the working sets of the applications in memory. Unfortunately, mobile devices often disable the invaluable swap feature as it can severely decrease the flash-based local storage device's already marginal lifespan due to the increased writes to the device. As a result, modern mobile devices suffering from the insufficient memory space end up killing memory-hungry applications and keeping only a few applications in the memory. In this paper, we propose CloudSwap, a fast and robust swap mechanism for mobile devices to enable the memory-oblivious application caching. The key idea of CloudSwap is to use the fast local storage as a cache of read-intensive swap pages, while storing prefetch-enabled, write-intensive swap pages in a cloud storage. To preserve the lifespan of the local storage, CloudSwap minimizes the number of writes to the local storage by storing the modified portions of the locally swapped pages in a cloud. To reduce the remote swap-in latency, CloudSwap exploits two cloud-assisted prefetch schemes, the app-aware read-ahead scheme and the access pattern-aware prefetch scheme. Our evaluation shows that the performance of CloudSwap is comparable to a fast, but lifespan-critical local swap system, with only 18% lifespan reduction, compared to the local swap system's 85% lifespan reduction.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"310 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132033790","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}
On current large-scale HPC platforms the data path from compute nodes to final storage passes through several networks interconnecting a distributed hierarchy of nodes serving as compute nodes, I/O nodes, and file system servers. Although applications compete for resources at various system levels, the current system software offers no mechanisms for globally coordinating the data flow for attaining optimal resource usage and for reacting to overload or interference. In this paper we describe CLARISSE, a middleware designed to enhance data-staging coordination and control in the HPC software storage I/O stack. CLARISSE exposes the parallel data flows to a higher-level hierarchy of controllers, thereby opening up the possibility of developing novel cross-layer optimizations, based on the run-time information. To the best of our knowledge, CLARISSE is the first middleware that decouples the policy, control, and data layers of the software I/O stack in order to simplify the task of globally coordinating the data staging on large-scale HPC platforms. To demonstrate how CLARISSE can be used for performance enhancement, we present two case studies: an elastic load-aware collective I/O and a cross-application parallel I/O scheduling policy. The evaluation illustrates how coordination can bring a significant performance benefit with low overheads by adapting to load conditions and interference.
{"title":"CLARISSE: A Middleware for Data-Staging Coordination and Control on Large-Scale HPC Platforms","authors":"Florin Isaila, J. Carretero, R. Ross","doi":"10.1109/CCGrid.2016.24","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.24","url":null,"abstract":"On current large-scale HPC platforms the data path from compute nodes to final storage passes through several networks interconnecting a distributed hierarchy of nodes serving as compute nodes, I/O nodes, and file system servers. Although applications compete for resources at various system levels, the current system software offers no mechanisms for globally coordinating the data flow for attaining optimal resource usage and for reacting to overload or interference. In this paper we describe CLARISSE, a middleware designed to enhance data-staging coordination and control in the HPC software storage I/O stack. CLARISSE exposes the parallel data flows to a higher-level hierarchy of controllers, thereby opening up the possibility of developing novel cross-layer optimizations, based on the run-time information. To the best of our knowledge, CLARISSE is the first middleware that decouples the policy, control, and data layers of the software I/O stack in order to simplify the task of globally coordinating the data staging on large-scale HPC platforms. To demonstrate how CLARISSE can be used for performance enhancement, we present two case studies: an elastic load-aware collective I/O and a cross-application parallel I/O scheduling policy. The evaluation illustrates how coordination can bring a significant performance benefit with low overheads by adapting to load conditions and interference.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134641837","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}
Ching-Hsiang Chu, Khaled Hamidouche, Akshay Venkatesh, A. Awan, D. Panda
Accelerators like NVIDIA GPUs have changed the landscape of current HPC clusters to a great extent. Massive heterogeneous parallelism offered by these accelerators have led to GPU-Aware MPI libraries that are widely used for writing distributed parallel scientific applications. Compute-oriented collective operations like MPI_Reduce perform computation on data in addition to the usual communication performed by collectives. Historically, these collectives, due to their compute requirements have been implemented on CPU (or Host) only. However, with the advent of GPU technologies it has become important for MPI libraries to provide better design for their GPU (or Device) based versions. In this paper, we tackle the above challenges and provide designs and implementations for most commonly used compute-oriented collectives - MPI_Reduce, MPI_Allreduce, and MPI_Scan - for GPU clusters. We propose extensions to the state-of-the-art algorithms to fully take advantage of the GPU capabilities like GPUDirect RDMA (GDR) and CUDA compute kernel to efficiently perform these operations. With our new designs, we report reduced execution time for all compute-based collectives up to 96 GPUs. Experimental results show an improvement of 50% for small messages and 85% for large messages using MPI_Reduce. For MPI_Allreduce and MPI_Scan, we report more than 40% reduction in time for large messages. Furthermore, analytical models are developed and evaluated to understand and predict the performance of proposed designs for extremely large-scale GPU clusters.
{"title":"CUDA Kernel Based Collective Reduction Operations on Large-scale GPU Clusters","authors":"Ching-Hsiang Chu, Khaled Hamidouche, Akshay Venkatesh, A. Awan, D. Panda","doi":"10.1109/CCGrid.2016.111","DOIUrl":"https://doi.org/10.1109/CCGrid.2016.111","url":null,"abstract":"Accelerators like NVIDIA GPUs have changed the landscape of current HPC clusters to a great extent. Massive heterogeneous parallelism offered by these accelerators have led to GPU-Aware MPI libraries that are widely used for writing distributed parallel scientific applications. Compute-oriented collective operations like MPI_Reduce perform computation on data in addition to the usual communication performed by collectives. Historically, these collectives, due to their compute requirements have been implemented on CPU (or Host) only. However, with the advent of GPU technologies it has become important for MPI libraries to provide better design for their GPU (or Device) based versions. In this paper, we tackle the above challenges and provide designs and implementations for most commonly used compute-oriented collectives - MPI_Reduce, MPI_Allreduce, and MPI_Scan - for GPU clusters. We propose extensions to the state-of-the-art algorithms to fully take advantage of the GPU capabilities like GPUDirect RDMA (GDR) and CUDA compute kernel to efficiently perform these operations. With our new designs, we report reduced execution time for all compute-based collectives up to 96 GPUs. Experimental results show an improvement of 50% for small messages and 85% for large messages using MPI_Reduce. For MPI_Allreduce and MPI_Scan, we report more than 40% reduction in time for large messages. Furthermore, analytical models are developed and evaluated to understand and predict the performance of proposed designs for extremely large-scale GPU clusters.","PeriodicalId":103641,"journal":{"name":"2016 16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid)","volume":"75 15","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134196959","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：