itransformer: Using SSD to Improve Disk
|
|
- Peregrine Singleton
- 3 years ago
- Views:
From this document you will learn the answers to the following questions:
What patterns can be found in a database?
What type of storage system is used for parallel file systems?
What is the main problem with using a queue?
Transcription
1 itransformer: Using SSD to Improve Disk Scheduling for High performance I/O Xuechen Zhang Song Jiang Wayne State t University it Kei Davis Los Alamos National Laboratory
2 Challenges of data management using parallel file systems Processing huge amount of data LANL combustion application generates 5 TB data per run. Checkpointing p gdatasets are 22 TB per timestep. Particle accelerator Large Hadron Collider at CERN generates over 15 PB of data per year. Handling very high I/O concurrency Intrepid p at Argonne has 164k cores. I/O intensive applications are concurrently executed. Increased I/O concurrency can compromise the efficiency of hard-disk-based parallel file systems, which are designed to handle huge amount of data. 2
3 Handling concurrency without disk scheduling Track number of incoming requests Disk service order w/o scheduler 1, 5, 7, 21, 3, 4, 2, 10 Seek distance 1, 5, 7, 21, 3, 4, 2,
4 Handling concurrency with I/O scheduler Track number of incoming requests 1, 5, 7, 21, 3, 4, 2, Scheduling of the first four requests Scheduling of the second four requests. Request service order: 1, 5, 7, 21, 2, 3, 4, 10 Using I/O scheduler (assuming queue size is 4): disk seek distance is 42. I/O scheduler helps reduce disk seek distance. 4
5 Handling concurrency with I/O scheduler of extended dispatchqueue Track number of incoming requests , 5, 7, 21, 3, 4, 2, Request service order: 1, 2, 3, 4, 5, 7, 10, 21 Using I/O scheduler (assuming queue size is 8): disk seek distance is 20. Increasing the queue length can further reduce the seek distance and help improve I/O efficiency.
6 Effect of queue size on I/O performance 1) Iozone a file system benchmark; 2) Various data a access patterns; 3) CFQ I/O scheduler; 4) 8GB data file; 5) 256 threads; 6 A large queue can effectively recover spatial locality if available in requests. When individual threads issue fully random requests, increasing queue size generates limited improvements.
7 Limitations of increasing queue size in disk scheduling Having a large queue would allow many write requests to be outstanding in volatile DRAM, running the risk of losing data. Keeping requests in the queue without being completed may increase requestresponse response times and violate QoS requirements. Increasing thequeuesize may not besufficient to address the issue of concurrency among streams of random requests. 7
8 Our solution: usingsolid state solid state disksto improve diskschedulingscheduling In-memory Queue This queue is only responsible for dispatching requests if relatively strong locality can be identified in the queue. itransformer In-SSD Queue Extension 1) Requests are sent to the queue extension on the SSD for further scheduling. 2) SSD is non-volatile, keeping dirty data in the queue extension safe. 3) A write request is considered complete once it arrives at the queue. 8
9 Talk outline Introduction Design and Implementation Performance evaluation Related work Conclusions 9
10 10 itransformer: a new disk schedulingframework usingsolid state solid state disks Design objectives Recover and exploit spatial locality lit in the I/O requests; Hide random access latency on the hard disk. The challenges How to effectively handle scientific workloads which have large working data sets and relatively weak locality? When scientific program processes a large data set, data accesses exhibit weak temporal locality, which is hard to exploit by a relatively small SSD for effective caching. When I/O concurrency is high, handling a large amount of random access can overwhelm SSD as a small storage device for random data. How to reduce overhead for multi level scheduling using both inmemory queue and in SSD queue extension? Serving sequential requests using SSD as buffer may increase I/O lt latency.
11 Scheduling modes Determining the scheduling modes MEM mode (in memory queue) When spatial locality is strong, only in memory queue is used for dispatching. Requests are directly sent from in memory queue to disk. SSD mode (in memory queue and in SSD queue extension) When spatial locality is weak, requests are sent from in memory queue to in SSD queue extension and then dispatched from SSD to disk. Weak Locality MEM mode SSD mode 11 Strong Locality
12 Locality measurement of dispatch queues 12 Locality measurement of requests from queues The average distance of disk head movement for serving requests To statistically quantify the spatial locality, we use the method which is similar to the one developed in Linux. Locality ofrequests fromin memoryqueue (Lin_mem) Locality of requests from in SSD queue extension (Lin_SSD) Locality estimation when SSD is not used When SSD is not in use, we still need an intelligent method to estimate the best locality which can be achieved if SSD had been used. itransformer maintains i a separate ghost queue to hold the metadata of any reqeusts dispatched out of the in memory queue. We run the standard disks scheduler over the requests in the ghost queue. Locality of requests from the ghost queue (Lghost) Comparing the localities li i to determine the transition ii of modes Lin_mem VS. Lin_SSD or Lin_mem VS. Lghost
13 Out of band writeback for random writes itransformer handles random write via out of band writeback to opportunistically reduce I/O latencies of random writes. When random writes are detected, dt tditransformer uses SSD to cache the data and delays the service of write requests as long as the SSD queue is not full. Write requests are later dispatched from SSDswhen hard disks are not busy. 13
14 Data prefeching for random reads Random reads Random reads are difficult to handle even when SSD is used. itransformer tracks history of random reads to predict and prefetch data to be requested into SSDs. Tracking I/O read activities Disk address space is partitioned into slots of equal size. A slot is the minimal unit for prefetching. We designed an algorithm to identify slots that random read requests have moved into and out of multiple l times. The slots which are randomly accessed in high frequency becomesthe candidates for prefethcing. 14
15 Illustration of the LRU stack S: slot number; C: access frequency; Slot Sequence: 3, 1, 1, 2 1, 1, 2 1, 2 2 Null Top S:1 C:1 S:3 C:2 S:1 C:2 S:1 C:2 S:2 C:1 S:9 C:1 S:1 C:1 S:3 C:2 S:3 C:2 S:1 C:2 S:5 C:1 S:9 C:1 S:9 C:1 S:9 C:1 S:3 C:2 S:3 C:1 S:5 C:1 S:5 C:1 S:5 C:1 S:9 C:1 When a slot s counter value is greater than a threshold, the slot becomes a prefetchable slot. 15
16 itransformer implementation itransformer is prototyped with instrumentation of the Linux device mapper as a stand alone kernel module in the generic block layer; To activate itransformer in a cluster system administrators only need load the module into the kernel. To maintain i the data consistency itransformer writes dirty data on the SSD back to disk on unloading of the module. During initializationitransformer itransformer checks if there are anydirty data left in the SSD because of system failure and rebuilds the mapping table. 16
17 Talk outline Introduction Design and Implementation Performance evaluation Related work Conclusions 17
18 Experimental setting Darwin cluster at Los Alamos National Laboratory 120 nodes and 8 data servers 48 core 2GHz AMD Opteron 6168, 64GB memory RAID0 consisting of two 500GB 7200rpm disks 120GB SSD Software configuration Kernel CFQ for RAID, NOOP for SSD MPICH compiled with ROMIO PVFS2 parallel file system 18
19 Comparison of storage devices SSD Hard disk RAID0 Capacity 120GB 1TB Interface SATA SAS Seq. Read 160MB/s 170MB/s Seq. Write 140MB/s 160MB/s Ran. Read 60MB/s 15MB/s Ran. Write 30MB/s 5MB/s 19
20 The ior mpi io benchmark Random access pattern Performance improvement up to 2.4X 20
21 On disk data access pattern Disk Disk SSD 21
22 The Hpio benchmark 22 Read Write 4.5X improvement on average 1) large number of random small writes 2) sync flush dirty data and meta-data t
23 The BTIO benchmark Performance increasingly suffers since request size is reduced to 200Bytes at 1024 processes. 23
24 The S3asim benchmark itransformer reduces I/O times by up to 66%. 24
25 Related work Hardware Solutions Use storage devices which are not sensitive to random access, e.g. solid state disk. It is not cost effective for scientific applications Large working set; Weak Spatial locality; Software strategies Adopt techniques inparallel I/Omiddleware, e.g. collective IO. Requires specific interfaces; It is not designed for multiple applications; Disk schedulers in Linux kernel, e.g. CFQ, NOOP, 25
26 Conclusions Existing schedulingmechanism makes disks vulnerable to high I/O concurrency. We propose itransformer a two level scheduling framework usingin SSD queueextension extension to handlei/o concurrency and randomness of workloads. Evaluation with representative parallel I/O benchmarks show itransformer increases I/O throughput by upto 3X and 35% on average. 26
27 Q & A Contacts: Xuechen Zhang, xczhang@wayne.edu Kei Davis, kei.davis@lanl.gov Song Jiang, sjiang@eng.wayne.edu 27
28 28 Performance impactof prefetch area size
29 Related work Middleware approaches to handlingi/o concurrency Collective I/O [Thakur et al. FRONTIERS 99] SSD based cache in thememoryhierarchy hierarchy Flashcache [Srinivasan et al. at Facebook] Readyboost [Microsoft] SSD based hybrid storage system Combo drive [Payer et al. 2009] I CASH [Ren et al. HPCA 11] Hystor [Chen et al. ICS 2011] 29
Opportunistic Data-driven Execution of Parallel Programs for Efficient I/O Services
Opportunistic Data-driven Execution of Parallel Programs for Efficient I/O Services Xuechen Zhang ECE Department Wayne State University Detroit, MI, 4822, USA xczhang@wayne.edu Kei Davis CCS Division Los
More informationPARALLELS CLOUD STORAGE
PARALLELS CLOUD STORAGE Performance Benchmark Results 1 Table of Contents Executive Summary... Error! Bookmark not defined. Architecture Overview... 3 Key Features... 5 No Special Hardware Requirements...
More informationUse of Hadoop File System for Nuclear Physics Analyses in STAR
1 Use of Hadoop File System for Nuclear Physics Analyses in STAR EVAN SANGALINE UC DAVIS Motivations 2 Data storage a key component of analysis requirements Transmission and storage across diverse resources
More informationChoosing Storage Systems
Choosing Storage Systems For MySQL Peter Zaitsev, CEO Percona Percona Live MySQL Conference and Expo 2013 Santa Clara,CA April 25,2013 Why Right Choice for Storage is Important? 2 because Wrong Choice
More informationInterferenceRemoval: Removing Interference of Disk Access for MPI Programs through Data Replication
InterferenceRemoval: Removing Interference of Disk Access for MPI Programs through Data Replication Xuechen Zhang and Song Jiang The ECE Department Wayne State University Detroit, MI, 4822, USA {xczhang,
More informationVirtuoso and Database Scalability
Virtuoso and Database Scalability By Orri Erling Table of Contents Abstract Metrics Results Transaction Throughput Initializing 40 warehouses Serial Read Test Conditions Analysis Working Set Effect of
More informationUsing Synology SSD Technology to Enhance System Performance Synology Inc.
Using Synology SSD Technology to Enhance System Performance Synology Inc. Synology_SSD_Cache_WP_ 20140512 Table of Contents Chapter 1: Enterprise Challenges and SSD Cache as Solution Enterprise Challenges...
More informationAccelerating Enterprise Applications and Reducing TCO with SanDisk ZetaScale Software
WHITEPAPER Accelerating Enterprise Applications and Reducing TCO with SanDisk ZetaScale Software SanDisk ZetaScale software unlocks the full benefits of flash for In-Memory Compute and NoSQL applications
More informationEvaluation Report: Accelerating SQL Server Database Performance with the Lenovo Storage S3200 SAN Array
Evaluation Report: Accelerating SQL Server Database Performance with the Lenovo Storage S3200 SAN Array Evaluation report prepared under contract with Lenovo Executive Summary Even with the price of flash
More informationViolin: A Framework for Extensible Block-level Storage
Violin: A Framework for Extensible Block-level Storage Michail Flouris Dept. of Computer Science, University of Toronto, Canada flouris@cs.toronto.edu Angelos Bilas ICS-FORTH & University of Crete, Greece
More informationCloud Storage. Parallels. Performance Benchmark Results. White Paper. www.parallels.com
Parallels Cloud Storage White Paper Performance Benchmark Results www.parallels.com Table of Contents Executive Summary... 3 Architecture Overview... 3 Key Features... 4 No Special Hardware Requirements...
More informationPerformance Characteristics of VMFS and RDM VMware ESX Server 3.0.1
Performance Study Performance Characteristics of and RDM VMware ESX Server 3.0.1 VMware ESX Server offers three choices for managing disk access in a virtual machine VMware Virtual Machine File System
More informationBoost SQL Server Performance Buffer Pool Extensions & Delayed Durability
Boost SQL Server Performance Buffer Pool Extensions & Delayed Durability Manohar Punna President - SQLServerGeeks #509 Brisbane 2016 Agenda SQL Server Memory Buffer Pool Extensions Delayed Durability Analysis
More informationA High-Performance Storage System for the LHCb Experiment Juan Manuel Caicedo Carvajal, Jean-Christophe Garnier, Niko Neufeld, and Rainer Schwemmer
658 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 57, NO. 2, APRIL 2010 A High-Performance Storage System for the LHCb Experiment Juan Manuel Caicedo Carvajal, Jean-Christophe Garnier, Niko Neufeld, and Rainer
More informationClient-aware Cloud Storage
Client-aware Cloud Storage Feng Chen Computer Science & Engineering Louisiana State University Michael Mesnier Circuits & Systems Research Intel Labs Scott Hahn Circuits & Systems Research Intel Labs Cloud
More informationManaging Storage Space in a Flash and Disk Hybrid Storage System
Managing Storage Space in a Flash and Disk Hybrid Storage System Xiaojian Wu, and A. L. Narasimha Reddy Dept. of Electrical and Computer Engineering Texas A&M University IEEE International Symposium on
More informationYouChoose: A Performance Interface Enabling Convenient and Efficient QoS Support for Consolidated Storage Systems
YouChoose: A Performance Interface Enabling Convenient and Efficient QoS Support for Consolidated Storage Systems Xuechen Zhang, Yuehai Xu, and Song Jiang Department of Electrical and Computer Engineering
More informationTaking Linux File and Storage Systems into the Future. Ric Wheeler Director Kernel File and Storage Team Red Hat, Incorporated
Taking Linux File and Storage Systems into the Future Ric Wheeler Director Kernel File and Storage Team Red Hat, Incorporated 1 Overview Going Bigger Going Faster Support for New Hardware Current Areas
More informationFlash-Friendly File System (F2FS)
Flash-Friendly File System (F2FS) Feb 22, 2013 Joo-Young Hwang (jooyoung.hwang@samsung.com) S/W Dev. Team, Memory Business, Samsung Electronics Co., Ltd. Agenda Introduction FTL Device Characteristics
More informationCloud Computing through Virtualization and HPC technologies
Cloud Computing through Virtualization and HPC technologies William Lu, Ph.D. 1 Agenda Cloud Computing & HPC A Case of HPC Implementation Application Performance in VM Summary 2 Cloud Computing & HPC HPC
More informationBenchmarking Cassandra on Violin
Technical White Paper Report Technical Report Benchmarking Cassandra on Violin Accelerating Cassandra Performance and Reducing Read Latency With Violin Memory Flash-based Storage Arrays Version 1.0 Abstract
More informationDatabase Hardware Selection Guidelines
Database Hardware Selection Guidelines BRUCE MOMJIAN Database servers have hardware requirements different from other infrastructure software, specifically unique demands on I/O and memory. This presentation
More informationWhy Computers Are Getting Slower (and what we can do about it) Rik van Riel Sr. Software Engineer, Red Hat
Why Computers Are Getting Slower (and what we can do about it) Rik van Riel Sr. Software Engineer, Red Hat Why Computers Are Getting Slower The traditional approach better performance Why computers are
More informationOverlapping Data Transfer With Application Execution on Clusters
Overlapping Data Transfer With Application Execution on Clusters Karen L. Reid and Michael Stumm reid@cs.toronto.edu stumm@eecg.toronto.edu Department of Computer Science Department of Electrical and Computer
More informationDirect NFS - Design considerations for next-gen NAS appliances optimized for database workloads Akshay Shah Gurmeet Goindi Oracle
Direct NFS - Design considerations for next-gen NAS appliances optimized for database workloads Akshay Shah Gurmeet Goindi Oracle Agenda Introduction Database Architecture Direct NFS Client NFS Server
More informationFlexible Storage Allocation
Flexible Storage Allocation A. L. Narasimha Reddy Department of Electrical and Computer Engineering Texas A & M University Students: Sukwoo Kang (now at IBM Almaden) John Garrison Outline Big Picture Part
More informationRAMCloud and the Low- Latency Datacenter. John Ousterhout Stanford University
RAMCloud and the Low- Latency Datacenter John Ousterhout Stanford University Most important driver for innovation in computer systems: Rise of the datacenter Phase 1: large scale Phase 2: low latency Introduction
More informationLSI MegaRAID FastPath Performance Evaluation in a Web Server Environment
LSI MegaRAID FastPath Performance Evaluation in a Web Server Environment Evaluation report prepared under contract with LSI Corporation Introduction Interest in solid-state storage (SSS) is high, and IT
More informationStorage benchmarking cookbook
Storage benchmarking cookbook How to perform solid storage performance measurements Stijn Eeckhaut Stijn De Smet, Brecht Vermeulen, Piet Demeester The situation today: storage systems can be very complex
More informationWHITE PAPER FUJITSU PRIMERGY SERVER BASICS OF DISK I/O PERFORMANCE
WHITE PAPER BASICS OF DISK I/O PERFORMANCE WHITE PAPER FUJITSU PRIMERGY SERVER BASICS OF DISK I/O PERFORMANCE This technical documentation is aimed at the persons responsible for the disk I/O performance
More informationSeptember 25, 2007. Maya Gokhale Georgia Institute of Technology
NAND Flash Storage for High Performance Computing Craig Ulmer cdulmer@sandia.gov September 25, 2007 Craig Ulmer Maya Gokhale Greg Diamos Michael Rewak SNL/CA, LLNL Georgia Institute of Technology University
More informationDIABLO TECHNOLOGIES MEMORY CHANNEL STORAGE AND VMWARE VIRTUAL SAN : VDI ACCELERATION
DIABLO TECHNOLOGIES MEMORY CHANNEL STORAGE AND VMWARE VIRTUAL SAN : VDI ACCELERATION A DIABLO WHITE PAPER AUGUST 2014 Ricky Trigalo Director of Business Development Virtualization, Diablo Technologies
More informationEnabling Enterprise Solid State Disks Performance
Carnegie Mellon University Research Showcase @ CMU Computer Science Department School of Computer Science 3-29 Enabling Enterprise Solid State Disks Performance Milo Polte Carnegie Mellon University Jiri
More informationIndexing on Solid State Drives based on Flash Memory
Indexing on Solid State Drives based on Flash Memory Florian Keusch MASTER S THESIS Systems Group Department of Computer Science ETH Zurich http://www.systems.ethz.ch/ September 2008 - March 2009 Supervised
More informationAIX NFS Client Performance Improvements for Databases on NAS
AIX NFS Client Performance Improvements for Databases on NAS October 20, 2005 Sanjay Gulabani Sr. Performance Engineer Network Appliance, Inc. gulabani@netapp.com Diane Flemming Advisory Software Engineer
More informationMulti-Threading Performance on Commodity Multi-Core Processors
Multi-Threading Performance on Commodity Multi-Core Processors Jie Chen and William Watson III Scientific Computing Group Jefferson Lab 12000 Jefferson Ave. Newport News, VA 23606 Organization Introduction
More informationEnhancing Recovery Using an SSD Buffer Pool Extension
Enhancing Recovery Using an SSD Buffer Pool Extension Bishwaranjan Bhattacharjee IBM T.J.Watson Research Center bhatta@us.ibm.com Christian Lang* Acelot Inc. clang@acelot.com George A Mihaila* Google Inc.
More informationBenchmarking Hadoop & HBase on Violin
Technical White Paper Report Technical Report Benchmarking Hadoop & HBase on Violin Harnessing Big Data Analytics at the Speed of Memory Version 1.0 Abstract The purpose of benchmarking is to show advantages
More informationBest Practices for Deploying SSDs in a Microsoft SQL Server 2008 OLTP Environment with Dell EqualLogic PS-Series Arrays
Best Practices for Deploying SSDs in a Microsoft SQL Server 2008 OLTP Environment with Dell EqualLogic PS-Series Arrays Database Solutions Engineering By Murali Krishnan.K Dell Product Group October 2009
More informationFile System & Device Drive. Overview of Mass Storage Structure. Moving head Disk Mechanism. HDD Pictures 11/13/2014. CS341: Operating System
CS341: Operating System Lect 36: 1 st Nov 2014 Dr. A. Sahu Dept of Comp. Sc. & Engg. Indian Institute of Technology Guwahati File System & Device Drive Mass Storage Disk Structure Disk Arm Scheduling RAID
More informationOutline. CS 245: Database System Principles. Notes 02: Hardware. Hardware DBMS ... ... Data Storage
CS 245: Database System Principles Notes 02: Hardware Hector Garcia-Molina Outline Hardware: Disks Access Times Solid State Drives Optimizations Other Topics: Storage costs Using secondary storage Disk
More informationCOS 318: Operating Systems
COS 318: Operating Systems File Performance and Reliability Andy Bavier Computer Science Department Princeton University http://www.cs.princeton.edu/courses/archive/fall10/cos318/ Topics File buffer cache
More informationComputer Engineering and Systems Group Electrical and Computer Engineering SCMFS: A File System for Storage Class Memory
SCMFS: A File System for Storage Class Memory Xiaojian Wu, Narasimha Reddy Texas A&M University What is SCM? Storage Class Memory Byte-addressable, like DRAM Non-volatile, persistent storage Example: Phase
More informationStoring Data: Disks and Files. Disks and Files. Why Not Store Everything in Main Memory? Chapter 7
Storing : Disks and Files Chapter 7 Yea, from the table of my memory I ll wipe away all trivial fond records. -- Shakespeare, Hamlet base Management Systems 3ed, R. Ramakrishnan and J. Gehrke 1 Disks and
More informationPerformance and scalability of a large OLTP workload
Performance and scalability of a large OLTP workload ii Performance and scalability of a large OLTP workload Contents Performance and scalability of a large OLTP workload with DB2 9 for System z on Linux..............
More informationPicking the right number of targets per server for BeeGFS. Jan Heichler March 2015 v1.2
Picking the right number of targets per server for BeeGFS Jan Heichler March 2015 v1.2 Evaluating the MetaData Performance of BeeGFS 2 Abstract In this paper we will show the performance of two different
More informationPerformance Analysis of Flash Storage Devices and their Application in High Performance Computing
Performance Analysis of Flash Storage Devices and their Application in High Performance Computing Nicholas J. Wright With contributions from R. Shane Canon, Neal M. Master, Matthew Andrews, and Jason Hick
More informationSolid State Storage in Massive Data Environments Erik Eyberg
Solid State Storage in Massive Data Environments Erik Eyberg Senior Analyst Texas Memory Systems, Inc. Agenda Taxonomy Performance Considerations Reliability Considerations Q&A Solid State Storage Taxonomy
More informationIDO: Intelligent Data Outsourcing with Improved RAID Reconstruction Performance in Large-Scale Data Centers
IDO: Intelligent Data Outsourcing with Improved RAID Reconstruction Performance in Large-Scale Data Centers Suzhen Wu 12, Hong Jiang 2, Bo Mao 2 1 Computer Science Department, Xiamen University 2 Department
More informationBENCHMARKING CLOUD DATABASES CASE STUDY on HBASE, HADOOP and CASSANDRA USING YCSB
BENCHMARKING CLOUD DATABASES CASE STUDY on HBASE, HADOOP and CASSANDRA USING YCSB Planet Size Data!? Gartner s 10 key IT trends for 2012 unstructured data will grow some 80% over the course of the next
More informationComparing Dynamic Disk Pools (DDP) with RAID-6 using IOR
Comparing Dynamic Disk Pools (DDP) with RAID-6 using IOR December, 2012 Peter McGonigal petermc@sgi.com Abstract Dynamic Disk Pools (DDP) offer an exciting new approach to traditional RAID sets by substantially
More informationGraySort on Apache Spark by Databricks
GraySort on Apache Spark by Databricks Reynold Xin, Parviz Deyhim, Ali Ghodsi, Xiangrui Meng, Matei Zaharia Databricks Inc. Apache Spark Sorting in Spark Overview Sorting Within a Partition Range Partitioner
More informationSQL Server Business Intelligence on HP ProLiant DL785 Server
SQL Server Business Intelligence on HP ProLiant DL785 Server By Ajay Goyal www.scalabilityexperts.com Mike Fitzner Hewlett Packard www.hp.com Recommendations presented in this document should be thoroughly
More informationAccelerating Server Storage Performance on Lenovo ThinkServer
Accelerating Server Storage Performance on Lenovo ThinkServer Lenovo Enterprise Product Group April 214 Copyright Lenovo 214 LENOVO PROVIDES THIS PUBLICATION AS IS WITHOUT WARRANTY OF ANY KIND, EITHER
More informationEnergy Efficient Storage Management Cooperated with Large Data Intensive Applications
Energy Efficient Storage Management Cooperated with Large Data Intensive Applications Norifumi Nishikawa #1, Miyuki Nakano #2, Masaru Kitsuregawa #3 # Institute of Industrial Science, The University of
More informationUsing Synology SSD Technology to Enhance System Performance. Based on DSM 5.2
Using Synology SSD Technology to Enhance System Performance Based on DSM 5.2 Table of Contents Chapter 1: Enterprise Challenges and SSD Cache as Solution Enterprise Challenges... 3 SSD Cache as Solution...
More informationReliability-Aware Energy Management for Hybrid Storage Systems
MSST Research Track, May 2011 Reliability-Aware Energy Management for Hybrid Storage Systems Wes Felter, Anthony Hylick, John Carter IBM Research - Austin Energy Saving using Hybrid Storage with Flash
More informationChapter 11 I/O Management and Disk Scheduling
Operating Systems: Internals and Design Principles, 6/E William Stallings Chapter 11 I/O Management and Disk Scheduling Dave Bremer Otago Polytechnic, NZ 2008, Prentice Hall I/O Devices Roadmap Organization
More informationHP Smart Array Controllers and basic RAID performance factors
Technical white paper HP Smart Array Controllers and basic RAID performance factors Technology brief Table of contents Abstract 2 Benefits of drive arrays 2 Factors that affect performance 2 HP Smart Array
More informationRecommended hardware system configurations for ANSYS users
Recommended hardware system configurations for ANSYS users The purpose of this document is to recommend system configurations that will deliver high performance for ANSYS users across the entire range
More informationIntel Solid- State Drive Data Center P3700 Series NVMe Hybrid Storage Performance
Intel Solid- State Drive Data Center P3700 Series NVMe Hybrid Storage Performance Hybrid Storage Performance Gains for IOPS and Bandwidth Utilizing Colfax Servers and Enmotus FuzeDrive Software NVMe Hybrid
More informationKVM PERFORMANCE IMPROVEMENTS AND OPTIMIZATIONS. Mark Wagner Principal SW Engineer, Red Hat August 14, 2011
KVM PERFORMANCE IMPROVEMENTS AND OPTIMIZATIONS Mark Wagner Principal SW Engineer, Red Hat August 14, 2011 1 Overview Discuss a range of topics about KVM performance How to improve out of the box experience
More informationFlash Performance in Storage Systems. Bill Moore Chief Engineer, Storage Systems Sun Microsystems
Flash Performance in Storage Systems Bill Moore Chief Engineer, Storage Systems Sun Microsystems 1 Disk to CPU Discontinuity Moore s Law is out-stripping disk drive performance (rotational speed) As a
More informationIn-Memory Databases Algorithms and Data Structures on Modern Hardware. Martin Faust David Schwalb Jens Krüger Jürgen Müller
In-Memory Databases Algorithms and Data Structures on Modern Hardware Martin Faust David Schwalb Jens Krüger Jürgen Müller The Free Lunch Is Over 2 Number of transistors per CPU increases Clock frequency
More informationSamsung Solid State Drive RAPID mode
Samsung Solid State Drive RAPID mode White Paper 2013 Samsung Electronics Co. Improving System Responsiveness with Samsung RAPID mode Innovative solution pairs advanced SSD technology with cutting-edge
More informationConfiguring RAID for Optimal Performance
Configuring RAID for Optimal Performance Intel RAID Controller SRCSASJV Intel RAID Controller SRCSASRB Intel RAID Controller SRCSASBB8I Intel RAID Controller SRCSASLS4I Intel RAID Controller SRCSATAWB
More informationEnabling Technologies for Distributed Computing
Enabling Technologies for Distributed Computing Dr. Sanjay P. Ahuja, Ph.D. Fidelity National Financial Distinguished Professor of CIS School of Computing, UNF Multi-core CPUs and Multithreading Technologies
More informationAvid ISIS 7000. www.avid.com
Avid ISIS 7000 www.avid.com Table of Contents Overview... 3 Avid ISIS Technology Overview... 6 ISIS Storage Blade... 6 ISIS Switch Blade... 7 ISIS System Director... 7 ISIS Client Software... 8 ISIS Redundant
More informationSummer Student Project Report
Summer Student Project Report Dimitris Kalimeris National and Kapodistrian University of Athens June September 2014 Abstract This report will outline two projects that were done as part of a three months
More informationConfiguring Apache Derby for Performance and Durability Olav Sandstå
Configuring Apache Derby for Performance and Durability Olav Sandstå Database Technology Group Sun Microsystems Trondheim, Norway Overview Background > Transactions, Failure Classes, Derby Architecture
More informationHigh Performance Computing Specialists. ZFS Storage as a Solution for Big Data and Flexibility
High Performance Computing Specialists ZFS Storage as a Solution for Big Data and Flexibility Introducing VA Technologies UK Based System Integrator Specialising in High Performance ZFS Storage Partner
More informationBottleneck Detection in Parallel File Systems with Trace-Based Performance Monitoring
Julian M. Kunkel - Euro-Par 2008 1/33 Bottleneck Detection in Parallel File Systems with Trace-Based Performance Monitoring Julian M. Kunkel Thomas Ludwig Institute for Computer Science Parallel and Distributed
More information89 Fifth Avenue, 7th Floor. New York, NY 10003. www.theedison.com 212.367.7400. White Paper. HP 3PAR Adaptive Flash Cache: A Competitive Comparison
89 Fifth Avenue, 7th Floor New York, NY 10003 www.theedison.com 212.367.7400 White Paper HP 3PAR Adaptive Flash Cache: A Competitive Comparison Printed in the United States of America Copyright 2014 Edison
More informationAirWave 7.7. Server Sizing Guide
AirWave 7.7 Server Sizing Guide Copyright 2013 Aruba Networks, Inc. Aruba Networks trademarks include, Aruba Networks, Aruba Wireless Networks, the registered Aruba the Mobile Edge Company logo, Aruba
More informationUsing Synology SSD Technology to Enhance System Performance Synology Inc.
Using Synology SSD Technology to Enhance System Performance Synology Inc. Synology_WP_ 20121112 Table of Contents Chapter 1: Enterprise Challenges and SSD Cache as Solution Enterprise Challenges... 3 SSD
More informationEnterprise Applications
Enterprise Applications Chi Ho Yue Sorav Bansal Shivnath Babu Amin Firoozshahian EE392C Emerging Applications Study Spring 2003 Functionality Online Transaction Processing (OLTP) Users/apps interacting
More informationSystem Architecture. CS143: Disks and Files. Magnetic disk vs SSD. Structure of a Platter CPU. Disk Controller...
System Architecture CS143: Disks and Files CPU Word (1B 64B) ~ 10 GB/sec Main Memory System Bus Disk Controller... Block (512B 50KB) ~ 100 MB/sec Disk 1 2 Magnetic disk vs SSD Magnetic Disk Stores data
More informationMaximizing VMware ESX Performance Through Defragmentation of Guest Systems. Presented by
Maximizing VMware ESX Performance Through Defragmentation of Guest Systems Presented by July, 2010 Table of Contents EXECUTIVE OVERVIEW 3 TEST EQUIPMENT AND METHODS 4 TESTING OVERVIEW 5 Fragmentation in
More informationSAP HANA - Main Memory Technology: A Challenge for Development of Business Applications. Jürgen Primsch, SAP AG July 2011
SAP HANA - Main Memory Technology: A Challenge for Development of Business Applications Jürgen Primsch, SAP AG July 2011 Why In-Memory? Information at the Speed of Thought Imagine access to business data,
More informationOn Benchmarking Popular File Systems
On Benchmarking Popular File Systems Matti Vanninen James Z. Wang Department of Computer Science Clemson University, Clemson, SC 2963 Emails: {mvannin, jzwang}@cs.clemson.edu Abstract In recent years,
More informationNew Cluster-Ready FAS3200 Models
New Cluster-Ready FAS3200 Models Steven Miller Senior Technical Director and Platform Architect NetApp recently introduced two new models in the FAS3200 series: the FAS3220 and the FAS3250. Our design
More informationExploiting Tier 0 and Virtualization to Maximize Storage Performance
Exploiting Tier 0 and Virtualization to Maximize Storage Performance Presented by Tim Conley (ATS Group) SaaS-built innovations by ATS to empower C-level management to IT administrators 2007-2011 ATS Group.
More informationCOS 318: Operating Systems. Virtual Machine Monitors
COS 318: Operating Systems Virtual Machine Monitors Andy Bavier Computer Science Department Princeton University http://www.cs.princeton.edu/courses/archive/fall10/cos318/ Introduction Have been around
More informationPetascale Software Challenges. Piyush Chaudhary piyushc@us.ibm.com High Performance Computing
Petascale Software Challenges Piyush Chaudhary piyushc@us.ibm.com High Performance Computing Fundamental Observations Applications are struggling to realize growth in sustained performance at scale Reasons
More informationA Virtual Storage Environment for SSDs and HDDs in Xen Hypervisor
A Virtual Storage Environment for SSDs and HDDs in Xen Hypervisor Yu-Jhang Cai, Chih-Kai Kang and Chin-Hsien Wu Department of Electronic and Computer Engineering National Taiwan University of Science and
More informationCommoditisation of the High-End Research Storage Market with the Dell MD3460 & Intel Enterprise Edition Lustre
Commoditisation of the High-End Research Storage Market with the Dell MD3460 & Intel Enterprise Edition Lustre University of Cambridge, UIS, HPC Service Authors: Wojciech Turek, Paul Calleja, John Taylor
More informationLab Evaluation of NetApp Hybrid Array with Flash Pool Technology
Lab Evaluation of NetApp Hybrid Array with Flash Pool Technology Evaluation report prepared under contract with NetApp Introduction As flash storage options proliferate and become accepted in the enterprise,
More informationIntroduction Disks RAID Tertiary storage. Mass Storage. CMSC 412, University of Maryland. Guest lecturer: David Hovemeyer.
Guest lecturer: David Hovemeyer November 15, 2004 The memory hierarchy Red = Level Access time Capacity Features Registers nanoseconds 100s of bytes fixed Cache nanoseconds 1-2 MB fixed RAM nanoseconds
More informationvpfs: Bandwidth Virtualization of Parallel Storage Systems
vpfs: Bandwidth Virtualization of Parallel Storage Systems Yiqi Xu, Dulcardo Arteaga, Ming Zhao Florida International University {yxu6,darte3,ming}@cs.fiu.edu Yonggang Liu, Renato Figueiredo University
More informationDavid Rioja Redondo Telecommunication Engineer Englobe Technologies and Systems
David Rioja Redondo Telecommunication Engineer Englobe Technologies and Systems About me David Rioja Redondo Telecommunication Engineer - Universidad de Alcalá >2 years building and managing clusters UPM
More informationRAID Performance Analysis
RAID Performance Analysis We have six 500 GB disks with 8 ms average seek time. They rotate at 7200 RPM and have a transfer rate of 20 MB/sec. The minimum unit of transfer to each disk is a 512 byte sector.
More informationMicrosoft Windows Server 2003 with Internet Information Services (IIS) 6.0 vs. Linux Competitive Web Server Performance Comparison
April 23 11 Aviation Parkway, Suite 4 Morrisville, NC 2756 919-38-28 Fax 919-38-2899 32 B Lakeside Drive Foster City, CA 9444 65-513-8 Fax 65-513-899 www.veritest.com info@veritest.com Microsoft Windows
More informationLow-Power Amdahl-Balanced Blades for Data-Intensive Computing
Thanks to NVIDIA, Microsoft External Research, NSF, Moore Foundation, OCZ Technology Low-Power Amdahl-Balanced Blades for Data-Intensive Computing Alex Szalay, Andreas Terzis, Alainna White, Howie Huang,
More informationCray DVS: Data Virtualization Service
Cray : Data Virtualization Service Stephen Sugiyama and David Wallace, Cray Inc. ABSTRACT: Cray, the Cray Data Virtualization Service, is a new capability being added to the XT software environment with
More informationParallels Cloud Server 6.0
Parallels Cloud Server 6.0 Parallels Cloud Storage I/O Benchmarking Guide September 05, 2014 Copyright 1999-2014 Parallels IP Holdings GmbH and its affiliates. All rights reserved. Parallels IP Holdings
More informationMemory Channel Storage ( M C S ) Demystified. Jerome McFarland
ory nel Storage ( M C S ) Demystified Jerome McFarland Principal Product Marketer AGENDA + INTRO AND ARCHITECTURE + PRODUCT DETAILS + APPLICATIONS THE COMPUTE-STORAGE DISCONNECT + Compute And Data Have
More informationGPU File System Encryption Kartik Kulkarni and Eugene Linkov
GPU File System Encryption Kartik Kulkarni and Eugene Linkov 5/10/2012 SUMMARY. We implemented a file system that encrypts and decrypts files. The implementation uses the AES algorithm computed through
More informationFacebook: Cassandra. Smruti R. Sarangi. Department of Computer Science Indian Institute of Technology New Delhi, India. Overview Design Evaluation
Facebook: Cassandra Smruti R. Sarangi Department of Computer Science Indian Institute of Technology New Delhi, India Smruti R. Sarangi Leader Election 1/24 Outline 1 2 3 Smruti R. Sarangi Leader Election
More informationHybrid Storage Management for Database Systems
Hybrid Storage Management for Database Systems Xin Liu University of Waterloo, Canada x39liu@uwaterloo.ca Kenneth Salem University of Waterloo, Canada ksalem@uwaterloo.ca ABSTRACT The use of flash-based
More informationIntel DPDK Boosts Server Appliance Performance White Paper
Intel DPDK Boosts Server Appliance Performance Intel DPDK Boosts Server Appliance Performance Introduction As network speeds increase to 40G and above, both in the enterprise and data center, the bottlenecks
More information