5,701 Hits in 4.1 sec

Symmetric active/active metadata service for high availability parallel file systems

Xubin He, Li Ou, Christian Engelmann, Xin Chen, Stephen L. Scott
2009 Journal of Parallel and Distributed Computing  
High availability data storage systems are critical for many applications as research and business become more data-driven.  ...  In addition, the replication overhead for multiple metadata services can be very high.  ...  IOzone generates and measures a variety of file operations for determining a broad file system analysis.  ... 
doi:10.1016/j.jpdc.2009.08.004 fatcat:j2nw7mjj2bchdgbqttgxm7rwzy

Parallel network file systems using authenticated key exchange protocols

S. Sathya, M. Ranjith Kumar, K. Madheswaran
2017 Journal of Applied and Advanced Research  
The keyestablishment for secure many-to-many communications is very important nowadays.  ...  This shows that these protocols are capable of reducing the workload of the metadata server and concurrently supporting forward secrecy and escrow-freeness.  ...  Existing and Proposed System Existing System Key establishment for secure many-to-many communications.  ... 
doi:10.21839/jaar.2017.v2i3.89 fatcat:3hdi7oxhpzd73fcnfref6ou6se

The Impact of Performance Asymmetry in Emerging Multicore Architectures

Saisanthosh Balakrishnan, Ravi Rajwar, Mike Upton, Konrad Lai
2005 SIGARCH Computer Architecture News  
In addition to asymmetry-aware operating system kernels, the application often itself needs to be aware of performance asymmetry for stable and scalable performance.  ...  Performance asymmetry adversely affects behavior of many workloads. We study ways to eliminate these effects.  ...  We thank Kai Ming Chan, Yen-Kuang Chen, Kingsum Chow, William Clifford, Kshitij Doshi, Shih-Lien Lu, and Hamesh Patel for assistance in setting up the workloads.  ... 
doi:10.1145/1080695.1070012 fatcat:7lng4gsp35dabcvhbwthubbsf4

Adaptive Middleware for Data Replication [chapter]

Jesús M. Milan-Franco, Ricardo Jiménez-Peris, Marta Patiño-Martínez, Bettina Kemme
2004 Lecture Notes in Computer Science  
An extensive performance analysis shows that the proposed adaptive replication solution can provide high throughput, good scalability, and low response times for changing loads and workloads with little  ...  In this paper, we present an adaptive distributed middleware for data replication that is able to adjust to changes in the amount of load submitted to the different replicas and to the type of workload  ...  Middle-R is responsible for concurrency and replica control. It uses a group communication system to disseminate information among the replicas.  ... 
doi:10.1007/978-3-540-30229-2_10 fatcat:kcsfmacrejeqrhjkgdyw2uzcce

Task optimization based on CPU pipeline technique in a multicore system

Bo Wang, Yongwei Wu, Weimin Zheng
2012 Computers and Mathematics with Applications  
Currently, the multicore system is prevalent in desktops, laptops or servers. The web proxy can save network traffic overhead and shorten communication cost.  ...  We propose the CP technique to build parallel tasks in a proxy system.  ...  The design for Linux system adopts a time-sharing mechanism to run many applications concurrently.  ... 
doi:10.1016/j.camwa.2011.09.024 fatcat:hqciaqwdibeqtb5cnj4j2vfipm

Empirical analysis of overheads in cluster environments

Brian K. Schmidt, Vaidy S. Sunderam
1994 Concurrency Practice and Experience  
While it is unrealistic to expect communication speeds comparable to hardware buses or switches, highspeed networks can provide sufficient capacity and performance for a variety of applications.  ...  Interestingly, we find that agenda parallelism and load balancing strategies contribute significantly more to better performance than improved communications or system tuning.  ...  Unfortunately, the algorithm assumes a symmetric communication pattern, but in practice the broadcast tree will become unbalanced in terms of workload and idle time.  ... 
doi:10.1002/cpe.4330060102 fatcat:zkhdwir7djcxpc2g7z2sfoxqt4

Analyzing the impact of system architecture on the scalability of OLTP engines for high-contention workloads

Raja Appuswamy, Angelos C. Anadiotis, Danica Porobic, Mustafa K. Iman, Anastasia Ailamaki
2017 Proceedings of the VLDB Endowment  
However, recent research has shown that even the state-of-the-art OLTP engines are unable to exploit available parallelism for high contention workloads.  ...  Using Trireme, we present an extensive experimental study to understand i) the impact of each system architecture on overall scalability, ii) the interaction between system architecture and concurrency  ...  ACKNOWLEDGMENTS We would like to thank Eliezer Levy, Shay Goikhman, the anonymous reviewers, and the DIAS lab members for their constructive feedback.  ... 
doi:10.14778/3149193.3149194 fatcat:cirxwunnifep3o6qny6rpmjcji


2007 Issues in Information Systems  
Experimental trials were run on various platforms: from 2 to 12 two processor units for MPI, and 8 and 16 units for HTTP.  ...  That communication was relayed via the MPI protocol using TCP packets with intensities in .0005 second range.  ...  Each host was configured with two 550 MHz CPUs that utilized symmetric multi-processing as supported by the Linux operating system.  ... 
doi:10.48009/2_iis_2007_367-371 fatcat:njqf3hipm5acvjfnhvqpj4yhbq

Cloud And Parallel Network File System Using Authenticated Key Exchange Protocols

2016 International Journal of Science and Research (IJSR)  
Already we studied the issues of key establishment for secure many-to-many communications.  ...  The system work focuses on the current Internet standard for such file systems, i.e., parallel Network File System (pNFS), which makes use of Kerberos to establish parallel session keys between clients  ...  This way, the workload of generating session keys is amortized over v for all the clients within the file system.  ... 
doi:10.21275/v5i5.nov163973 fatcat:zem74bm3wvbtjmvdsvey2egx4a

The energy case for graph processing on hybrid CPU and GPU systems

Abdullah Gharaibeh, Elizeu Santos-Neto, Lauro Beltrão Costa, Matei Ripeanu
2013 Proceedings of the 3rd Workshop on Irregular Applications Architectures and Algorithms - IA^3 '13  
., CPU and GPU) single-node systems.  ...  Graph processing can be accelerated on hybrid systems by properly mapping the graphlayout to processing units, such that the algorithmic tasks exercise each of the units where they perform best.  ...  Apart from the BFS benchmark on UK-WEB graph, a hybrid 1S1G system improves both raw performance and power-normalized performance compared to the symmetric 2S system.  ... 
doi:10.1145/2535753.2535755 dblp:conf/sc/GharaibehSCR13 fatcat:pisruzaig5bg3k3dgayxbn2hca

Software Model-based Performance Analysis [chapter]

Dorina C. Petriu
2013 Model-Driven Engineering for Distributed Real-Time Systems  
The chapter ends with a discussion of lesson learned and future challenges for integrating the analysis of multiple non-functional properties in the context of MDE.  ...  The principles for transforming annotated software models into performance models are presented next.  ...  , 2005) and the UML Profile for Modeling and Analysis of Real-Time and Embedded systems (MARTE) defined for UML2.X versions (OMG, 2009) .  ... 
doi:10.1002/9781118558096.ch6 fatcat:b4vo3xaoijhzhjdjitem3wftoa

An Empirical-cum-Statistical Approach to Power-Performance Characterization of Concurrent GPU Kernels [article]

Nilanjan Goswami, Amer Qouneh, Chao Li, Tao Li
2020 arXiv   pre-print
Also, we demonstrate a multi-kernel throughput benchmark suite based on the framework that encapsulates symmetric, asymmetric and co-existing (often appears together) kernel based workloads.  ...  Besides, we explain several findings regarding power-performance co-optimization of concurrent throughput workloads.  ...  (b) We also propose a systematic methodology for multikernel throughput workload creation for emerging systems.  ... 
arXiv:2011.02368v2 fatcat:xgce6gvcjjcilfwem452yd3hsi

Consistent Data Replication: Is It Feasible in WANs? [chapter]

Yi Lin, Bettina Kemme, Marta Patiño-Martínez, Ricardo Jiménez-Peris
2005 Lecture Notes in Computer Science  
Our experiments show that performance remains acceptable even for medium sized systems, and data replication guaranteeing 1-copy-serializability is a serious alternative to weaker approaches in WAN environments  ...  As before, a request for transaction © is multicast in total order to all sites and the locks requested according to its total order delivery.  ...  Fig. 9 shows the response times for update transactions. A centralized system is always the best for updates since there is no communication.  ... 
doi:10.1007/11549468_70 fatcat:53ptdocdibhhbh56kqj4ftjrje

HiPC 2021 Workshop on Parallel Programming in the Exascale Era (PPEE 2021)

Vivek Kumar, Swarnendu Biswas, Vishwesh Jatala
2021 2021 IEEE 28th International Conference on High Performance Computing, Data and Analytics Workshop (HiPCW)  
The four critical challenges for exascale systems are extreme parallelism, power demand, data movement, and reliability.  ...  Failures on these systems are anticipated to occur many times a day, such that the existing approach for resiliency, such as checkpointing and restart, will not work.  ...  For FA-BSP, we extend classical actors with multiple symmetric mailboxes for scalability, and with automatic termination detection of messages initiated in a superstep.  ... 
doi:10.1109/hipcw54834.2021.00014 fatcat:pgtv7sjr7jewnjtgxg5j36n4li

Scale-out ccNUMA

Vasilis Gavrielatos, Antonios Katsarakis, Arpit Joshi, Nicolai Oswald, Boris Grot, Vijay Nagarajan
2018 Proceedings of the Thirteenth EuroSys Conference on - EuroSys '18  
We introduce symmetric caching, wherein every server node is provisioned with a small cache that maintains the most popular objects in the dataset.  ...  Despite recent work on skew mitigation, existing approaches offer only limited benefit for high-throughput in-memory KVS deployments.  ...  Trigonakis for his support on the concurrency control and Stephan Diestelhorst, Aleksandar Dragojevic, our shepherd, Kimberly Keeton, and the anonymous reviewers for their valuable comments.  ... 
doi:10.1145/3190508.3190550 dblp:conf/eurosys/GavrielatosKJOG18 fatcat:k5u2nhpfhbeopj5lmbdgbznlyu
« Previous Showing results 1 — 15 out of 5,701 results