Locally repairable codes
|
|
- Melissa Rose
- 5 years ago
- Views:
Transcription
1 Locally repairable codes for large-scale storage Presented by Anwitaman Datta Nanyang Technological University, Singapore Joint work with Frédérique Oggier & Lluís Pàmies i IISc 2012 & NetApp A. Datta, NTU Bangalore, SingaporeOctober 2012
2 Who am I? //sands ntu sg/ 2
3 S*Aspects of Networked Distributed Systems recommendation and decision support systems decentralized online social networking and collaboration Applica ations privacy aware/preserved data aggregation, storage, sharing & analytics/data-mining data/computation at 3 rd party/outsourced distributed key-value stores data-center design P2P/F2F storage systems networked distributed storage & data management systems (Distributed d) Systems social network analysis trust secure/privacy models preserved computation primitives codes for storage Founda ational
4 Large-scale storage: Disclaimer A note from the trenches: "You know you have a large storage system when you get paged at 1 AM because you only have a few petabytes of storage left." from Andrew Fikes (Principal Engineer, Google) faculty summit talk ` Storage Architecture and Challenges `, and some ask/say: why do you care about efficient storage space utilization, it is so cheap... I never get such calls!! 4
5 5 Source: data-center-expansion-plans
6 Scale how? To scale vertically (or scale up) means to add resources to a single node in a system* To scale horizontally (or scale out) means to add more nodes to a system, such as adding a new computer to a distributed software application* 6 Scale up Scale out * Definitions from Wikipedia
7 Distribution is essential Scaling up May just not even be feasible Even if feasible, it will be very expensive What happens when the machine fails? Scaling out => distributed storage Distribution => added complexity and vulnerabilities latency, consistency, faults, CAP theorem Consistency, Availability, Partition tolerance choose any two? but, not distributing is not a choice! 7
8 Failure Is Inevitable But, failure of the system is not an option! Failure is the pillar of rivals success Solution: Redundancy & Distribution 8
9 Five Levels of Redundancy Physical Virtual resource Availability zone Region Cloud From: 9
10 Redundancy Based Fault Tolerance Replicate data e.g., 3 or more copies In nodes on different racks Can deal with switch failures Power back-up using battery between racks (Google) 10
11 But At What Cost? Failure is not an option, but are the overheads acceptable? 11
12 Reducing the Overheads of Redundancy Erasure codes Much lower storage overhead High level of fault-tolerance In contrast to replication or RAID based systems Has the potential to significantly improve the bottomline e.g., Both Google s new DFS Collossus, as well as Microsoft s Azure now use ECs 12
13 Erasure Codes (ECs) An (n,k) erasure code = a map that takes as input k blocks and outputs n blocks, thus introducing n-k blocks of redundancy. 3 way replication is a (3,1) erasure code! Encoding k=1 block n=3 encoded blocks An erasure code such that the k original blocks can be recreated out of any k encoded blocks is called MDS (maximum distance separable). 13
14 Reed-Solomon Codes (named after Irving S. Reed and Gustave Solomon) Reed-Solomon Codes are well-known erasure codes. Encoding of (o 1,,o k ) is done by polynomial evaluation: The encoding blocks are then p(α 1 ),,p(α n ). 14
15 Erasure Codes (ECs) Originally designed for communication EC(n,k) Data = message O 1 O 2 O k Encoding B 1 Receive any k ( k) blocks B 2 O 1 Decoding Lost blocks O 2 Reconstru uct Data k blocks B n n encoded blocks O k Original k blocks 15
16 Erasure Codes for Networked Storage Data = Ob bject O 1 O 2 B 2 Encoding B 1 Retrieve any O k ( k) blocks B l Decoding O 1 O 2 Re econstruct t Data O k k blocks B n Lost blocks n encoded blocks O k Oii Original k blocks (stored in storage devices in a network) 16
17 Static Resilience Replicated r times Faults that can be tolerated: r-1 Probability of failure: f r Storage efficiency: 1/r Access: Find any one good replica Erasure coded (k of n) Faults that can be tolerated: n-k Probability of failure: k n n k j f f j n k j (1 ) 1 Storage efficiency: k/n Access: Find k good blocks Assumption: Peer failure is iid i.i.d. with failure probability f 17 k j replic ca object replic ca object For f=0.1 its 10-3 replic ca For f=0.1 3 of 9 code Blk Blk Blk Blk Blk Blk its ~3*10-6 Blk Blk Blk
18 Replenishing Lost Redundancy for ECs B 1 B 2 Repair needed for long term resilience. Retrieve any k ( k) blocks Decoding O 1 O 2 Encoding Recreate lost blocks B l Re-insert B n Lost blocks n encoded blocks 18 O k Original k blocks Repairs are expensive! Reinsert in (new) storage devices, so that there is (again) n encoded blocks
19 CanWeHaveBetterRepairability? Erasure codes tailor-made for distributed networked storage. Localized repairs: E.g., Hierarchical & Pyramid codes Locally repairable codes: E.g., Self-repairing codes, Punctured RM 19 Note: Network coding inspired regenerating codes also aim for better repairability, and specifically to minimize repair bandwidth, however our focus here are codes that reduce the repair fan-in (reduction of bandwidth is often an additional benefit.)
20 What to Tailored-Make the Codes for? Desired code properties include: Low storage overhead Good fault tolerance Better repairability Better Traditional MDS erasure codes achieve these. Smaller repair fan-in Reduced I/O for repairs Possibility of multiple simultaneous repairs Fast repairs Efficient B/W usage Better data-insertion Better migration to archival 20
21 B an Localized repairs: Hierarchical Codes Ạ bottom-up approach Essentially nested use of erasure codes a1 O B Bl l a1 O a2 B l an subgroup global redundancy from the code groups Replicate code group B g 1 local global local redundancy redundancy redundancy B g r B l s1 O s1 O s2 B l sn subgroup Code group Multi-hierarchical extension 21
22 Hierarchical Codes If `small number of faults Communication restricted within the `hierarchy suffice Progressively go higher-up for larger number of faults Isolated faults can be repaired independently Naturally maps to hierarchical data-center design? Asymmetry Different encoded blocks have different importance Difficult to analyze Complex algorithm (for decoding/repair) and system design Pros Cons Hierarchical Codes: How to Make Erasure Codes Attractive for Peer to Peer Storage Systems A. Duminico, E. P2P
23 Localized repairs: Pyramid Codes A top-down approach Example: Consider a MDS (11,8) code A (12,8) Pyramid code derived from the above MDS code: where 23
24 Pyramid Codes Good degraded read performance Cheaper repairs Fault-tolerance: There are regimes with deterministic behavior, and some intermediate regimes with probabilistic behavior So, easier to understand and reason about the system An optimized i version (called d`local reconstruction code ) )is used in Microsoft s Azure system Can be readily extended into a multi-tier pyramid Pyramid Codes: Flexible Schemes to Trade Space for Access Efficiency in Reliable Data Storage Systems Cheng Huang, Minghua Chen, and Jin NCA
25 Locally Repairable Codes The name is reminiscent of locally decodable codes Codes satisfying: encoded d fragments can be repaired directly from other small subsets (<< k) of encoded fragments Achievable also by codes supporting localized repairs (sometimes) number of live nodes contacted for repair (i.e., fan-in) is infact minimized Fan-in being some small constant, such as 2 or 3 typically independent of code parameters n & k a fragment can be repaired from a fixed number of encoded fragments, independently of which specific blocks are missing Analogous to erasure codes supporting reconstruction using any n - k losses, independently d of which h Partly achieved by some codes supporting localized repairs 25
26 Homomorphic Self-repairing Codes (HSRC) Usual disclaimer: To the best of our knowledge First instance of a locally repairable code Since then, there have been other instances, including Note another SRC variant we proposed (using projective geometric construction - PSRC) from other groups, e.g., punctured Reed-Muller codes k encoded blocks are enough to recreate the object Caveat: not any arbitrary k (i.e., SRCs are not MDS) However, there are many such k combinations Self repairing Homomorphic Codes for Distributed Storage Systems Frédérique Oggier and Anwitaman Datta Infocom
27 Self-repairing Codes: Blackbox View B 1 Retrieve some B 2 k (< k) blocks (e.g. k =2) to recreate a lost block B l Re-insert B n Lost blocks n encoded blocks (stored in storage devices in a network) Reinsert in (new) storage devices, so that there is (again) n encoded blocks 27
28 Homomorphic Self-repairing Codes (HSRC) Preliminaries 28
29 HSRC encoding 29
30 Self-repairing Codes Data = Object B O 1 W/ Linearized polynomial 1 B 2 O 2 Encoding with B l O k k blocks (Each of size M/k) B n n encoded blocks There is at least one pair to repair a node, for up to (n-1)/2 simultaneous failures (Parallel & fast repair of multiple faults) 30
31 HSRC(7,3) example 31
32 HSRC(7,3) example 32
33 HSRC(15,4) example: fast repair Consider Possible pairs to repair each block One possible parallelized repair schedule 33
34 PSRC Example Self repairing Codes for Distributed Storage Systems A Projective Geometric Construction FrédériqueOggier and AnwitamanDatta ITW 2011 (o 1 +o 2 +o 4 ) + (o 1 ) => o 2 +o 4 Repair using two nodes (o 3 ) + (o 2 +o 3 ) => o 2 Say N (o 1 ) + (o 2 ) => o 1 + o 1 and N 3 2 Four pieces needed to regenerate two pieces 34 Repair using three nodes Say N 2, N 3 and N 4 (o 2 ) + (o 4 ) => o 2 + o 4 (o 1 +o 2 +o 4 )+(o)=>o 4 ) o 1 +o 2 Three pieces needed 2012 A. to Datta, regenerate NTU two Singapore pieces
35 PSRC Example: Reconstruction o 3 o 4 (o 3 ) + (o 1 +o 3 ) => o 1 (o 1 )+(o 4 )+(o 1 +o 2 +o 4 ) => o 2 Reconstruction, say using N 3, N 4 and N 5 35
36 Maximum Distance Separable (MDS)? SRC is not MDS (and can not be!) Does it matter? Not much In practice, access will be planned PSRC(21,3) This is with random access 36
37 Practical properties (Current) SRCs are not systematic PSRC is like systematic Need to contact more nodes (than k) To obtain systematic `pieces Same total bandwidth usage Parallel a download oad for access can even be an `advantage `mixed strategies for access, i.e. get some systematic pieces, and some others Power saving (by switching off nodes) strategies possible? Coding/decoding gin PSRC are both using XOR operations only 37
38 Some very recent stuff Data insertion In-network coding for opportunistic back-up (arxiv: ) based on HSRC, exploiting the dependency of encoded pieces Generalization of the idea (for arbitrary LRCs) is outstanding Migrating replicated data into erasure encoded archived storage RapidRAID (arxiv: ) Has some local repairability properties, but that aspect is yet to be explored Another code ICDCN 2013 Systematic code (unlike RapidRAID) Found using numerical methods, and a general theory for the construction of such codes, as well as their repairability properties are open issues 38
39 The data insertion problem 39 For erasure coded data storage Traditionally centralized: Source node needs to encode and store at all storage nodes In contrast: replication is amenable to pipelining Further, if source/recipients are not available at the same time Two issues Storage nodes are busy with other tasks P2P/F2F settings: recipient nodes are offline Partly decentralizing the encoding process Leveraging on SRC s local repairability property Determining a good scheduling mechanism based on nodes availability Even with global (and future) knowledge, computationally prohibitive Scheduling heuristics Summary of results (simulations, so take with a grain of salt) Up to 90% speed-up in storage throughput with Google data center availability and workload traces Up to 60% speed-up with a F2F systems availability trace
40 The data migration to archival problem Data is often not accessed after a little while in the system, and thus can be archived using a storage-efficient efficient erasure code Speeding up this conversion to archival Decentralizing the encoding process Exploiting the existing replication of the object RapidRAID: Summary of results (on a proprietary 50 node cluster of HP ThinClients, and on EC2 instances) Up to 90% reduction of coding time of a single object Up to 20% reduction of coding time for batch processes Implementation (by Lluís Pàmies i Juárez) is available at: 40
41 Future steps/wishlist How to adapt different data management and analytics techniques to be compatible with erasure encoded data? Marrying deduplication techniques with erasure coding techniques Support for mutable content MTTDL analysis for codes with better repairability A complete working system that can be used out of the box by end users HDFS compatible 41
42 Outlook Interested to Follow: sce ntu edu sg/codingfornetworkedstorage/ Two surveys on storage codes one short, at high level, another longer with technical details Get involved: 42
On Coding Techniques for Networked Distributed Storage Systems
On Coding Techniques for Networked Distributed Storage Systems Frédérique Oggier frederique@ntu.edu.sg Nanyang Technological University, Singapore First European Training School on Network Coding, Barcelona,
More informationErasure Codes for Heterogeneous Networked Storage Systems
Erasure Codes for Heterogeneous Networked Storage Systems Lluís Pàmies i Juárez Lluís Pàmies i Juárez lpjuarez@ntu.edu.sg . Introduction Outline 2. Distributed Storage Allocation Problem 3. Homogeneous
More informationModern Erasure Codes for Distributed Storage Systems
Modern Erasure Codes for Distributed Storage Systems Storage Developer Conference, SNIA, Bangalore Srinivasan Narayanamurthy Advanced Technology Group, NetApp May 27 th 2016 1 Everything around us is changing!
More informationModern Erasure Codes for Distributed Storage Systems
Modern Erasure Codes for Distributed Storage Systems Srinivasan Narayanamurthy (Srini) NetApp Everything around us is changing! r The Data Deluge r Disk capacities and densities are increasing faster than
More informationRepair Pipelining for Erasure-Coded Storage
Repair Pipelining for Erasure-Coded Storage Runhui Li, Xiaolu Li, Patrick P. C. Lee, Qun Huang The Chinese University of Hong Kong USENIX ATC 2017 1 Introduction Fault tolerance for distributed storage
More informationScalable Data Management and Storage
Scalable Data Management and Storage on the Cloud: - State of the Art and Emerging Trends Presented by Anwitaman Datta & Frédérique Oggier NTU Singapore ICDCN 2012, Hong Kong Who are we? http://sands.sce.ntu.edu.sg/
More informationActiveScale Erasure Coding and Self Protecting Technologies
WHITE PAPER AUGUST 2018 ActiveScale Erasure Coding and Self Protecting Technologies BitSpread Erasure Coding and BitDynamics Data Integrity and Repair Technologies within The ActiveScale Object Storage
More informationPerformance Models of Access Latency in Cloud Storage Systems
Performance Models of Access Latency in Cloud Storage Systems Qiqi Shuai Email: qqshuai@eee.hku.hk Victor O.K. Li, Fellow, IEEE Email: vli@eee.hku.hk Yixuan Zhu Email: yxzhu@eee.hku.hk Abstract Access
More informationResiliency at Scale in the Distributed Storage Cloud
Resiliency at Scale in the Distributed Storage Cloud Alma Riska Advanced Storage Division EMC Corporation In collaboration with many at Cloud Infrastructure Group Outline Wi topic but this talk will focus
More informationGiza: Erasure Coding Objects across Global Data Centers
Giza: Erasure Coding Objects across Global Data Centers Yu Lin Chen*, Shuai Mu, Jinyang Li, Cheng Huang *, Jin li *, Aaron Ogus *, and Douglas Phillips* New York University, *Microsoft Corporation USENIX
More informationFast Erasure Coding for Data Storage: A Comprehensive Study of the Acceleration Techniques. Tianli Zhou & Chao Tian Texas A&M University
Fast Erasure Coding for Data Storage: A Comprehensive Study of the Acceleration Techniques Tianli Zhou & Chao Tian Texas A&M University 2 Contents Motivation Background and Review Evaluating Individual
More informationActiveScale Erasure Coding and Self Protecting Technologies
NOVEMBER 2017 ActiveScale Erasure Coding and Self Protecting Technologies BitSpread Erasure Coding and BitDynamics Data Integrity and Repair Technologies within The ActiveScale Object Storage System Software
More informationCloud Computing and Hadoop Distributed File System. UCSB CS170, Spring 2018
Cloud Computing and Hadoop Distributed File System UCSB CS70, Spring 08 Cluster Computing Motivations Large-scale data processing on clusters Scan 000 TB on node @ 00 MB/s = days Scan on 000-node cluster
More informationOptimize Storage Efficiency & Performance with Erasure Coding Hardware Offload. Dror Goldenberg VP Software Architecture Mellanox Technologies
Optimize Storage Efficiency & Performance with Erasure Coding Hardware Offload Dror Goldenberg VP Software Architecture Mellanox Technologies SNIA Legal Notice The material contained in this tutorial is
More informationHandling Big Data an overview of mass storage technologies
SS Data & Handling Big Data an overview of mass storage technologies Łukasz Janyst CERN IT Department CH-1211 Genève 23 Switzerland www.cern.ch/it GridKA School 2013 Karlsruhe, 26.08.2013 What is Big Data?
More informationHadoop File System S L I D E S M O D I F I E D F R O M P R E S E N T A T I O N B Y B. R A M A M U R T H Y 11/15/2017
Hadoop File System 1 S L I D E S M O D I F I E D F R O M P R E S E N T A T I O N B Y B. R A M A M U R T H Y Moving Computation is Cheaper than Moving Data Motivation: Big Data! What is BigData? - Google
More informationSoftware-defined Storage: Fast, Safe and Efficient
Software-defined Storage: Fast, Safe and Efficient TRY NOW Thanks to Blockchain and Intel Intelligent Storage Acceleration Library Every piece of data is required to be stored somewhere. We all know about
More informationLoad rebalancing with advanced encryption standard for Hadoop Distributed File System using Distributed Hash table
Volume 119 No. 12 2018, 13807-13814 ISSN: 1314-3395 (on-line version) url: http://www.ijpam.eu ijpam.eu Load rebalancing with advanced encryption standard for Hadoop Distributed File System using Distributed
More informationDistributed Systems. Characteristics of Distributed Systems. Lecture Notes 1 Basic Concepts. Operating Systems. Anand Tripathi
1 Lecture Notes 1 Basic Concepts Anand Tripathi CSci 8980 Operating Systems Anand Tripathi CSci 8980 1 Distributed Systems A set of computers (hosts or nodes) connected through a communication network.
More informationDistributed Systems. Characteristics of Distributed Systems. Characteristics of Distributed Systems. Goals in Distributed System Designs
1 Anand Tripathi CSci 8980 Operating Systems Lecture Notes 1 Basic Concepts Distributed Systems A set of computers (hosts or nodes) connected through a communication network. Nodes may have different speeds
More informationPrivacy, Security, and Repair in Distributed Storage Systems. Siddhartha Kumar
Privacy, Security, and Repair in Distributed Storage Systems Siddhartha Kumar Thesis for the Degree of Philosophiae Doctor (PhD) University of Bergen, Norway 2018 Privacy, Security, and Repair in Distributed
More informationA Survey on Erasure Coding Techniques for Cloud Storage System
A Survey on Erasure Coding Techniques for Cloud Storage System Kiran P. Pawar #1, R. M. Jogdand *2 #1 M.Tech, Department of Computer Science, Gogte Institute of technology, Udyambag Belagavi, Karnataka,
More informationCooperative Pipelined Regeneration in Distributed Storage Systems
Cooperative ipelined Regeneration in Distributed Storage Systems Jun Li, in Wang School of Computer Science Fudan University, China Baochun Li Department of Electrical and Computer Engineering University
More informationHuge market -- essentially all high performance databases work this way
11/5/2017 Lecture 16 -- Parallel & Distributed Databases Parallel/distributed databases: goal provide exactly the same API (SQL) and abstractions (relational tables), but partition data across a bunch
More informationDecentralized Distributed Storage System for Big Data
Decentralized Distributed Storage System for Big Presenter: Wei Xie -Intensive Scalable Computing Laboratory(DISCL) Computer Science Department Texas Tech University Outline Trends in Big and Cloud Storage
More informationDistributed Data Infrastructures, Fall 2017, Chapter 2. Jussi Kangasharju
Distributed Data Infrastructures, Fall 2017, Chapter 2 Jussi Kangasharju Chapter Outline Warehouse-scale computing overview Workloads and software infrastructure Failures and repairs Note: Term Warehouse-scale
More informationDistributed File Systems II
Distributed File Systems II To do q Very-large scale: Google FS, Hadoop FS, BigTable q Next time: Naming things GFS A radically new environment NFS, etc. Independence Small Scale Variety of workloads Cooperation
More informationDistributed Filesystem
Distributed Filesystem 1 How do we get data to the workers? NAS Compute Nodes SAN 2 Distributing Code! Don t move data to workers move workers to the data! - Store data on the local disks of nodes in the
More informationAssignment 5. Georgia Koloniari
Assignment 5 Georgia Koloniari 2. "Peer-to-Peer Computing" 1. What is the definition of a p2p system given by the authors in sec 1? Compare it with at least one of the definitions surveyed in the last
More informationHADOOP 3.0 is here! Dr. Sandeep Deshmukh Sadepach Labs Pvt. Ltd. - Let us grow together!
HADOOP 3.0 is here! Dr. Sandeep Deshmukh sandeep@sadepach.com Sadepach Labs Pvt. Ltd. - Let us grow together! About me BE from VNIT Nagpur, MTech+PhD from IIT Bombay Worked with Persistent Systems - Life
More informationCapacity Assurance in Hostile Networks
PhD Dissertation Defense Wednesday, October 7, 2015 3:30 pm - 5:30 pm 3112 Engineering Building Capacity Assurance in Hostile Networks By: Jian Li Advisor: Jian Ren ABSTRACT Linear network coding provides
More informationOn Data Parallelism of Erasure Coding in Distributed Storage Systems
On Data Parallelism of Erasure Coding in Distributed Storage Systems Jun Li, Baochun Li Department of Electrical and Computer Engineering, University of Toronto, Canada {junli, bli}@ece.toronto.edu Abstract
More informationBuilding High Speed Erasure Coding Libraries for ARM and x86 Processors. Per Simonsen, CEO, MemoScale May 2017
Building High Speed Erasure Coding Libraries for ARM and x86 Processors Per Simonsen, CEO, MemoScale May 2017 Agenda MemoScale company and team Erasure coding - brief intro MemoScale erasure codes Performance
More informationYour Data is in the Cloud: Who Exactly is Looking After It?
Your Data is in the Cloud: Who Exactly is Looking After It? P Vijay Kumar Dept of Electrical Communication Engineering Indian Institute of Science IISc Open Day March 4, 2017 1/33 Your Data is in the Cloud:
More informationCPSC 426/526. Cloud Computing. Ennan Zhai. Computer Science Department Yale University
CPSC 426/526 Cloud Computing Ennan Zhai Computer Science Department Yale University Recall: Lec-7 In the lec-7, I talked about: - P2P vs Enterprise control - Firewall - NATs - Software defined network
More informationAn Information-Theoretic Perspective of Consistent Distributed Storage
An Information-Theoretic Perspective of Consistent Distributed Storage Viveck R. Cadambe Pennsylvania State University Joint with Prof. Zhiying Wang (UCI) Prof. Nancy Lynch (MIT), Prof. Muriel Medard (MIT)
More informationAn Empirical Study of the Repair Performance of Novel Coding Schemes for Networked Distributed Storage Systems
An Empirical Study of the Repair Performance of Novel Coding Schemes for Networked Distributed Storage Systems Lluis Pamies-Juarez, Frédérique Oggier, and Anwitaman Datta 2 arxiv:26.287v [cs.dc] Jun 22
More informationCONFIGURATION GUIDE WHITE PAPER JULY ActiveScale. Family Configuration Guide
WHITE PAPER JULY 2018 ActiveScale Family Configuration Guide Introduction The world is awash in a sea of data. Unstructured data from our mobile devices, emails, social media, clickstreams, log files,
More informationEnabling Efficient and Reliable Transition from Replication to Erasure Coding for Clustered File Systems
Enabling Efficient and Reliable Transition from Replication to Erasure Coding for Clustered File Systems Runhui Li, Yuchong Hu, Patrick P. C. Lee Department of Computer Science and Engineering, The Chinese
More informationEight Tips for Better Archives. Eight Ways Cloudian Object Storage Benefits Archiving with Veritas Enterprise Vault
Eight Tips for Better Email Archives Eight Ways Cloudian Object Storage Benefits Email Archiving with Veritas Enterprise Vault Most organizations now manage terabytes, if not petabytes, of corporate and
More informationHDFS Architecture. Gregory Kesden, CSE-291 (Storage Systems) Fall 2017
HDFS Architecture Gregory Kesden, CSE-291 (Storage Systems) Fall 2017 Based Upon: http://hadoop.apache.org/docs/r3.0.0-alpha1/hadoopproject-dist/hadoop-hdfs/hdfsdesign.html Assumptions At scale, hardware
More informationCodes for Modern Applications
Codes for Modern Applications Current Research Topics in the Code and Signal Design Group P. Vijay Kumar Indian Institute of Science, Bangalore, India Nov. 16, 2018 1/28 Codes for Modern Applications:
More informationNetwork Coding for Distributed Storage Systems* Presented by Jayant Apte ASPITRG 7/9/13 & 7/11/13
Network Coding for Distributed Storage Systems* Presented by Jayant Apte ASPITRG 7/9/13 & 7/11/13 *Dimakis, A.G.; Godfrey, P.B.; Wu, Y.; Wainwright, M.J.; Ramchandran, K. "Network Coding for Distributed
More informationThe Construction of Open Source Cloud Storage System for Digital Resources
2017 3rd International Conference on Electronic Information Technology and Intellectualization (ICEITI 2017) ISBN: 978-1-60595-512-4 The Construction of Open Source Cloud Storage System for Digital Resources
More informationECS High Availability Design
ECS High Availability Design March 2018 A Dell EMC white paper Revisions Date Mar 2018 Aug 2017 July 2017 Description Version 1.2 - Updated to include ECS version 3.2 content Version 1.1 - Updated to include
More informationWrite a technical report Present your results Write a workshop/conference paper (optional) Could be a real system, simulation and/or theoretical
Identify a problem Review approaches to the problem Propose a novel approach to the problem Define, design, prototype an implementation to evaluate your approach Could be a real system, simulation and/or
More informationStaggeringly Large Filesystems
Staggeringly Large Filesystems Evan Danaher CS 6410 - October 27, 2009 Outline 1 Large Filesystems 2 GFS 3 Pond Outline 1 Large Filesystems 2 GFS 3 Pond Internet Scale Web 2.0 GFS Thousands of machines
More informationWindows Servers In Microsoft Azure
$6/Month Windows Servers In Microsoft Azure What I m Going Over 1. How inexpensive servers in Microsoft Azure are 2. How I get Windows servers for $6/month 3. Why Azure hosted servers are way better 4.
More informationData Integrity Protection scheme for Minimum Storage Regenerating Codes in Multiple Cloud Environment
International Journal of Advanced Research in Computer Engineering & Technology (IJARCET) Data Integrity Protection scheme for Minimum Storage Regenerating Codes in Multiple Cloud Environment Saravana
More informationSecurely Access Services Over AWS PrivateLink. January 2019
Securely Access Services Over AWS PrivateLink January 2019 Notices This document is provided for informational purposes only. It represents AWS s current product offerings and practices as of the date
More informationOn the Speedup of Single-Disk Failure Recovery in XOR-Coded Storage Systems: Theory and Practice
On the Speedup of Single-Disk Failure Recovery in XOR-Coded Storage Systems: Theory and Practice Yunfeng Zhu, Patrick P. C. Lee, Yuchong Hu, Liping Xiang, and Yinlong Xu University of Science and Technology
More informationEnabling Node Repair in Any Erasure Code for Distributed Storage
Enabling Node Repair in Any Erasure Code for Distributed Storage K. V. Rashmi, Nihar B. Shah, and P. Vijay Kumar, Fellow, IEEE Abstract Erasure codes are an efficient means of storing data across a network
More informationGUIDE. Optimal Network Designs with Cohesity
Optimal Network Designs with Cohesity TABLE OF CONTENTS Introduction...3 Key Concepts...4 Five Common Configurations...5 3.1 Simple Topology...5 3.2 Standard Topology...6 3.3 Layered Topology...7 3.4 Cisco
More informationResearch Faculty Summit Systems Fueling future disruptions
Research Faculty Summit 2018 Systems Fueling future disruptions Elevating the Edge to be a Peer of the Cloud Kishore Ramachandran Embedded Pervasive Lab, Georgia Tech August 2, 2018 Acknowledgements Enrique
More informationAgenda. AWS Database Services Traditional vs AWS Data services model Amazon RDS Redshift DynamoDB ElastiCache
Databases on AWS 2017 Amazon Web Services, Inc. and its affiliates. All rights served. May not be copied, modified, or distributed in whole or in part without the express consent of Amazon Web Services,
More informationIEEE 2013 JAVA PROJECTS Contact No: KNOWLEDGE AND DATA ENGINEERING
IEEE 2013 JAVA PROJECTS www.chennaisunday.com Contact No: 9566137117 KNOWLEDGE AND DATA ENGINEERING (DATA MINING) 1. A Fast Clustering-Based Feature Subset Selection Algorithm for High Dimensional Data
More informationOptimal Exact-Regenerating Codes for Distributed Storage at the MSR and MBR Points via a Product-Matrix Construction
IEEE TRANSACTIONS ON INFORMATION THEORY, VOL 57, NO 8, AUGUST 2011 5227 Optimal Exact-Regenerating Codes for Distributed Storage at the MSR and MBR Points via a Product-Matrix Construction K V Rashmi,
More informationECE Enterprise Storage Architecture. Fall 2018
ECE590-03 Enterprise Storage Architecture Fall 2018 RAID Tyler Bletsch Duke University Slides include material from Vince Freeh (NCSU) A case for redundant arrays of inexpensive disks Circa late 80s..
More informationFederated Array of Bricks Y Saito et al HP Labs. CS 6464 Presented by Avinash Kulkarni
Federated Array of Bricks Y Saito et al HP Labs CS 6464 Presented by Avinash Kulkarni Agenda Motivation Current Approaches FAB Design Protocols, Implementation, Optimizations Evaluation SSDs in enterprise
More informationChallenges in the Wide-area. Tapestry: Decentralized Routing and Location. Global Computation Model. Cluster-based Applications
Challenges in the Wide-area Tapestry: Decentralized Routing and Location System Seminar S 0 Ben Y. Zhao CS Division, U. C. Berkeley Trends: Exponential growth in CPU, b/w, storage Network expanding in
More informationPARALLEL CONSENSUS PROTOCOL
CANOPUS: A SCALABLE AND MASSIVELY PARALLEL CONSENSUS PROTOCOL Bernard Wong CoNEXT 2017 Joint work with Sajjad Rizvi and Srinivasan Keshav CONSENSUS PROBLEM Agreement between a set of nodes in the presence
More informationDesigning Fault-Tolerant Applications
Designing Fault-Tolerant Applications Miles Ward Enterprise Solutions Architect Building Fault-Tolerant Applications on AWS White paper published last year Sharing best practices We d like to hear your
More informationProgramming Systems for Big Data
Programming Systems for Big Data CS315B Lecture 17 Including material from Kunle Olukotun Prof. Aiken CS 315B Lecture 17 1 Big Data We ve focused on parallel programming for computational science There
More informationCloud Storage Reliability for Big Data Applications: A State of the Art Survey
Cloud Storage Reliability for Big Data Applications: A State of the Art Survey Rekha Nachiappan a, Bahman Javadi a, Rodrigo Calherios a, Kenan Matawie a a School of Computing, Engineering and Mathematics,
More informationEverything You Wanted To Know About Storage (But Were Too Proud To Ask) The Basics
Everything You Wanted To Know About Storage (But Were Too Proud To Ask) The Basics Today s Presenters Bob Plumridge HDS Chief Technology Officer - EMEA Alex McDonald NetApp CTO Office 2 SNIA Legal Notice
More informationEnabling Data Integrity Protection in Regenerating-Coding-Based Cloud Storage: Theory and Implementation (Supplementary File)
1 Enabling Data Integrity Protection in Regenerating-Coding-Based Cloud Storage: Theory and Implementation (Supplementary File) Henry C. H. Chen and Patrick P. C. Lee 1 ADDITIONAL RELATED WORK This section
More informationDistributed Systems. Lec 10: Distributed File Systems GFS. Slide acks: Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung
Distributed Systems Lec 10: Distributed File Systems GFS Slide acks: Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung 1 Distributed File Systems NFS AFS GFS Some themes in these classes: Workload-oriented
More informationNutanix Tech Note. Virtualizing Microsoft Applications on Web-Scale Infrastructure
Nutanix Tech Note Virtualizing Microsoft Applications on Web-Scale Infrastructure The increase in virtualization of critical applications has brought significant attention to compute and storage infrastructure.
More informationThe Microsoft Large Mailbox Vision
WHITE PAPER The Microsoft Large Mailbox Vision Giving users large mailboxes without breaking your budget Introduction Giving your users the ability to store more email has many advantages. Large mailboxes
More informationApache Hadoop 3. Balazs Gaspar Sales Engineer CEE & CIS Cloudera, Inc. All rights reserved.
Apache Hadoop 3 Balazs Gaspar Sales Engineer CEE & CIS balazs@cloudera.com 1 We believe data can make what is impossible today, possible tomorrow 2 We empower people to transform complex data into clear
More informationSelf-Adaptive Two-Dimensional RAID Arrays
Self-Adaptive Two-Dimensional RAID Arrays Jehan-François Pâris 1 Dept. of Computer Science University of Houston Houston, T 77204-3010 paris@cs.uh.edu Thomas J. E. Schwarz Dept. of Computer Engineering
More informationDisclaimer This presentation may contain product features that are currently under development. This overview of new technology represents no commitme
STO1926BU A Day in the Life of a VSAN I/O Diving in to the I/O Flow of vsan John Nicholson (@lost_signal) Pete Koehler (@vmpete) VMworld 2017 Content: Not for publication #VMworld #STO1926BU Disclaimer
More informationScaling Distributed Machine Learning
Scaling Distributed Machine Learning with System and Algorithm Co-design Mu Li Thesis Defense CSD, CMU Feb 2nd, 2017 nx min w f i (w) Distributed systems i=1 Large scale optimization methods Large-scale
More informationSCALING UP OF E-MSR CODES BASED DISTRIBUTED STORAGE SYSTEMS WITH FIXED NUMBER OF REDUNDANCY NODES
SCALING UP OF E-MSR CODES BASED DISTRIBUTED STORAGE SYSTEMS WITH FIXED NUMBER OF REDUNDANCY NODES Haotian Zhao, Yinlong Xu and Liping Xiang School of Computer Science and Technology, University of Science
More informationTake Back Lost Revenue by Activating Virtuozzo Storage Today
Take Back Lost Revenue by Activating Virtuozzo Storage Today JUNE, 2017 2017 Virtuozzo. All rights reserved. 1 Introduction New software-defined storage (SDS) solutions are enabling hosting companies to
More informationInternational Journal of Innovations in Engineering and Technology (IJIET)
RTL Design and Implementation of Erasure Code for RAID system Chethan.K 1, Dr.Srividya.P 2, Mr.Sivashanmugam Krishnan 3 1 PG Student, Department Of ECE, R. V. College Engineering, Bangalore, India. 2 Associate
More informationSCALABLE CONSISTENCY AND TRANSACTION MODELS
Data Management in the Cloud SCALABLE CONSISTENCY AND TRANSACTION MODELS 69 Brewer s Conjecture Three properties that are desirable and expected from realworld shared-data systems C: data consistency A:
More informationGFS Overview. Design goals/priorities Design for big-data workloads Huge files, mostly appends, concurrency, huge bandwidth Design for failures
GFS Overview Design goals/priorities Design for big-data workloads Huge files, mostly appends, concurrency, huge bandwidth Design for failures Interface: non-posix New op: record appends (atomicity matters,
More informationPRESENTATION TITLE GOES HERE. Understanding Architectural Trade-offs in Object Storage Technologies
Object Storage 201 PRESENTATION TITLE GOES HERE Understanding Architectural Trade-offs in Object Storage Technologies SNIA Legal Notice The material contained in this tutorial is copyrighted by the SNIA
More informationThe Technology Behind Datrium Cloud DVX
The Technology Behind Datrium Cloud DVX 385 Moffett Park Dr. Sunnyvale, CA 94089 844-478-8349 www.datrium.com Technical Report Public cloud as a new backup target Dedicated tape or disk based backup and
More informationCLUSTERING HIVEMQ. Building highly available, horizontally scalable MQTT Broker Clusters
CLUSTERING HIVEMQ Building highly available, horizontally scalable MQTT Broker Clusters 12/2016 About this document MQTT is based on a publish/subscribe architecture that decouples MQTT clients and uses
More informationProgramming Models MapReduce
Programming Models MapReduce Majd Sakr, Garth Gibson, Greg Ganger, Raja Sambasivan 15-719/18-847b Advanced Cloud Computing Fall 2013 Sep 23, 2013 1 MapReduce In a Nutshell MapReduce incorporates two phases
More informationTake control of storage performance
Take control of storage performance Transition From Speed To Management SSD + RAID 2008-2011 Reduce time to market Inherent bottlenecks Re-architect for better performance NVMe, SCSI Express Reads & Writes
More informationDiskReduce: Making Room for More Data on DISCs. Wittawat Tantisiriroj
DiskReduce: Making Room for More Data on DISCs Wittawat Tantisiriroj Lin Xiao, Bin Fan, and Garth Gibson PARALLEL DATA LABORATORY Carnegie Mellon University GFS/HDFS Triplication GFS & HDFS triplicate
More informationWHITEPAPER. MemSQL Enterprise Feature List
WHITEPAPER MemSQL Enterprise Feature List 2017 MemSQL Enterprise Feature List DEPLOYMENT Provision and deploy MemSQL anywhere according to your desired cluster configuration. On-Premises: Maximize infrastructure
More informationPyramid Codes: Flexible Schemes to Trade Space for Access Efficiency in Reliable Data Storage Systems
Pyramid Codes: Flexible Schemes to Trade Space for Access Efficiency in Reliable Data Storage Systems Cheng Huang, Minghua Chen, and Jin Li Microsoft Research, Redmond, WA 98052 Abstract To flexibly explore
More informationHYDRAstor: a Scalable Secondary Storage
HYDRAstor: a Scalable Secondary Storage 7th TF-Storage Meeting September 9 th 00 Łukasz Heldt Largest Japanese IT company $4 Billion in annual revenue 4,000 staff www.nec.com Polish R&D company 50 engineers
More informationDatabase Management Systems
Database Management Systems Distributed Databases Doug Shook What does it mean to be distributed? Multiple nodes connected by a network Data on the nodes is logically related The nodes do not need to be
More informationCourse Content. Parallel & Distributed Databases. Objectives of Lecture 12 Parallel and Distributed Databases
Database Management Systems Winter 2003 CMPUT 391: Dr. Osmar R. Zaïane University of Alberta Chapter 22 of Textbook Course Content Introduction Database Design Theory Query Processing and Optimisation
More informationChapter 18: Parallel Databases
Chapter 18: Parallel Databases Introduction Parallel machines are becoming quite common and affordable Prices of microprocessors, memory and disks have dropped sharply Recent desktop computers feature
More informationArchitekturen für die Cloud
Architekturen für die Cloud Eberhard Wolff Architecture & Technology Manager adesso AG 08.06.11 What is Cloud? National Institute for Standards and Technology (NIST) Definition On-demand self-service >
More informationMapReduce: Simplified Data Processing on Large Clusters 유연일민철기
MapReduce: Simplified Data Processing on Large Clusters 유연일민철기 Introduction MapReduce is a programming model and an associated implementation for processing and generating large data set with parallel,
More informationEC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures
EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures Haiyang Shi, Xiaoyi Lu, and Dhabaleswar K. (DK) Panda {shi.876, lu.932, panda.2}@osu.edu The Ohio State University
More informationEfficient Load Balancing and Disk Failure Avoidance Approach Using Restful Web Services
Efficient Load Balancing and Disk Failure Avoidance Approach Using Restful Web Services Neha Shiraz, Dr. Parikshit N. Mahalle Persuing M.E, Department of Computer Engineering, Smt. Kashibai Navale College
More informationKaminario Powering Cloud-Scale Applications with All-Flash. Sundip Arora Director, Product Marketing
Kaminario Powering Cloud-Scale Applications with All-Flash Sundip Arora Director, Product Marketing 275+ Employees Boston HQ Locations in Israel, London, Paris & Seoul 200+ Channel Partners $218M Total
More informationElastic Cloud Storage (ECS)
Elastic Cloud Storage (ECS) Version 3.1 Administration Guide 302-003-863 02 Copyright 2013-2017 Dell Inc. or its subsidiaries. All rights reserved. Published September 2017 Dell believes the information
More informationRAID SEMINAR REPORT /09/2004 Asha.P.M NO: 612 S7 ECE
RAID SEMINAR REPORT 2004 Submitted on: Submitted by: 24/09/2004 Asha.P.M NO: 612 S7 ECE CONTENTS 1. Introduction 1 2. The array and RAID controller concept 2 2.1. Mirroring 3 2.2. Parity 5 2.3. Error correcting
More informationDifferentiating Your Datacentre in the Networked Future John Duffin
Differentiating Your Datacentre in the Networked Future John Duffin Managing Director, South Asia. Uptime Institute July 2017 2017 Uptime Institute, LLC The Global Datacentre Authority 2 2017 Uptime Institute,
More informationManaging IoT and Time Series Data with Amazon ElastiCache for Redis
Managing IoT and Time Series Data with ElastiCache for Redis Darin Briskman, ElastiCache Developer Outreach Michael Labib, Specialist Solutions Architect 2016, Web Services, Inc. or its Affiliates. All
More informationNetworked Systems and Services, Fall 2018 Chapter 2. Jussi Kangasharju Markku Kojo Lea Kutvonen
Networked Systems and Services, Fall 2018 Chapter 2 Jussi Kangasharju Markku Kojo Lea Kutvonen Outline Physical layer reliability Low level reliability Parities and checksums Cyclic Redundancy Check (CRC)
More information