Kinetic Open Storage Platform: Enabling Break-through Economics in Scale-out Object Storage PRESENTATION TITLE GOES HERE Ali Fenn & James Hughes

Similar documents
Seagate Kinetic Open Storage Platform. Mayur Shetty - Senior Solutions Architect

Kinetic drive. Bingzhe Li

At-Scale Data Centers & Demand for New Architectures

Storage for HPC, HPDA and Machine Learning (ML)

Quobyte The Data Center File System QUOBYTE INC.

November 7, DAN WILSON Global Operations Architecture, Concur. OpenStack Summit Hong Kong JOE ARNOLD

Database Architecture 2 & Storage. Instructor: Matei Zaharia cs245.stanford.edu

Introducing SUSE Enterprise Storage 5

Deep Storage for Exponential Data. Nathan Thompson CEO, Spectra Logic

SMORE: A Cold Data Object Store for SMR Drives

IBM Spectrum NAS, IBM Spectrum Scale and IBM Cloud Object Storage

GlusterFS Architecture & Roadmap

Hadoop An Overview. - Socrates CCDH

SCS Distributed File System Service Proposal

EMC ISILON HARDWARE PLATFORM

Scality RING on Cisco UCS: Store File, Object, and OpenStack Data at Scale

Cold Storage: The Road to Enterprise Ilya Kuznetsov YADRO

Why software defined storage matters? Sergey Goncharov Solution Architect, Red Hat

An Exploration into Object Storage for Exascale Supercomputers. Raghu Chandrasekar

Emerging Technologies for HPC Storage

CEPH APPLIANCE Take a leap into the next generation of enterprise storage

Data Analytics and Storage System (DASS) Mixing POSIX and Hadoop Architectures. 13 November 2016

IP-Based Object Drives Now Have a Management Standard

UNIFY DATA AT MEMORY SPEED. Haoyuan (HY) Li, Alluxio Inc. VAULT Conference 2017

Increase Value from Big Data with Real-Time Data Integration and Streaming Analytics

BIG DATA READY WITH ISILON JEUDI 19 NOVEMBRE Bertrand OUNANIAN: Advisory System Engineer

EsgynDB Enterprise 2.0 Platform Reference Architecture

How Apache Hadoop Complements Existing BI Systems. Dr. Amr Awadallah Founder, CTO Cloudera,

HDFS: Hadoop Distributed File System. Sector: Distributed Storage System

Cloud Computing and Hadoop Distributed File System. UCSB CS170, Spring 2018

BeoLink.org. Design and build an inexpensive DFS. Fabrizio Manfredi Furuholmen. FrOSCon August 2008

Ceph Intro & Architectural Overview. Abbas Bangash Intercloud Systems

Nowcasting. D B M G Data Base and Data Mining Group of Politecnico di Torino. Big Data: Hype or Hallelujah? Big data hype?

Crossing the Chasm: Sneaking a parallel file system into Hadoop

Topics. Big Data Analytics What is and Why Hadoop? Comparison to other technologies Hadoop architecture Hadoop ecosystem Hadoop usage examples

AN ALTERNATIVE TO ALL- FLASH ARRAYS: PREDICTIVE STORAGE CACHING

BIG DATA AND HADOOP ON THE ZFS STORAGE APPLIANCE

Data Movement & Tiering with DMF 7

RAIDIX Data Storage Solution. Clustered Data Storage Based on the RAIDIX Software and GPFS File System

New Fresh Storage Approach for New IT Challenges Laurent Denel Philippe Nicolas OpenIO

Deploying Software Defined Storage for the Enterprise with Ceph. PRESENTATION TITLE GOES HERE Paul von Stamwitz Fujitsu

Distributed Storage with GlusterFS

Next Generation Storage for The Software-Defned World

TECHNICAL OVERVIEW OF NEW AND IMPROVED FEATURES OF EMC ISILON ONEFS 7.1.1

Scale-Out backups with Bareos and Gluster. Niels de Vos Gluster co-maintainer Red Hat Storage Developer

DataDirect Networks steps up object storage push with WOS refresh

Flash Storage with 24G SAS Leads the Way in Crunching Big Data

Enterprise Architectures The Pace Accelerates Camberley Bates Managing Partner & Analyst

HPE Storage news. Mauro Colombo Hybrid IT sales & presales manager 23 rd May 2018

INTRODUCTION TO CEPH. Orit Wasserman Red Hat August Penguin 2017

TITLE: PRE-REQUISITE THEORY. 1. Introduction to Hadoop. 2. Cluster. Implement sort algorithm and run it using HADOOP

VMware Virtual SAN Technology

BIG DATA STRATEGY FOR TODAY AND TOMORROW RYAN SAYRE, EMC ISILON CTO-AT-LARGE, EMEA

BUSINESS DATA LAKE FADI FAKHOURI, SR. SYSTEMS ENGINEER, ISILON SPECIALIST. Copyright 2016 EMC Corporation. All rights reserved.

Big Data in OpenStack Storage

Crossing the Chasm: Sneaking a parallel file system into Hadoop

Combine Native SQL Flexibility with SAP HANA Platform Performance and Tools

Backtesting with Spark

HPE Synergy HPE SimpliVity 380

CA485 Ray Walshe Google File System

DEMYSTIFYING BIG DATA WITH RIAK USE CASES. Martin Schneider Basho Technologies!

5 Fundamental Strategies for Building a Data-centered Data Center

NVMFS: A New File System Designed Specifically to Take Advantage of Nonvolatile Memory

Functional Testing of SQL Server on Kaminario K2 Storage

Introduction to Scientific Data Management

Introduction to Scientific Data Management

Provisioning with SUSE Enterprise Storage. Nyers Gábor Trainer &

IBM Storwize V7000, Storwize V5000 and IBM Storwize V5000F

New Approach to Unstructured Data

Universal Storage. Innovation to Break Decades of Tradeoffs VASTDATA.COM

Warehouse- Scale Computing and the BDAS Stack

Modernize Your Storage

Software Defined Storage for the Evolving Data Center

MODERNISE WITH ALL-FLASH. Intel Inside. Powerful Data Centre Outside.

Data Protection for Cisco HyperFlex with Veeam Availability Suite. Solution Overview Cisco Public

Stages of Data Processing

A BigData Tour HDFS, Ceph and MapReduce

Evaluating Cloud Storage Strategies. James Bottomley; CTO, Server Virtualization

Cisco UCS B440 M1High-Performance Blade Server

Isilon: Raising The Bar On Performance & Archive Use Cases. John Har Solutions Product Manager Unstructured Data Storage Team

DDN s Vision for the Future of Lustre LUG2015 Robert Triendl

SwiftStack and python-swiftclient

朱义普. Resolving High Performance Computing and Big Data Application Bottlenecks with Application-Defined Flash Acceleration. Director, North Asia, HPC

SAS Technical Update Connectivity Roadmap and MultiLink SAS Initiative Jay Neer Molex Corporation Marty Czekalski Seagate Technology LLC

Design a Remote-Office or Branch-Office Data Center with Cisco UCS Mini

Cisco and Cloudera Deliver WorldClass Solutions for Powering the Enterprise Data Hub alerts, etc. Organizations need the right technology and infrastr

Software Development Using Full System Simulation with Freescale QorIQ Communications Processors

HCI: Hyper-Converged Infrastructure

Webinar Series: Triangulate your Storage Architecture with SvSAN Caching. Luke Pruen Technical Services Director

A product by CloudFounders. Wim Provoost Open vstorage

2014 年 3 月 13 日星期四. From Big Data to Big Value Infrastructure Needs and Huawei Best Practice

How to solve your backup problems with HP StoreOnce

Software Defined Storage

A New Key-value Data Store For Heterogeneous Storage Architecture Intel APAC R&D Ltd.

System that permanently stores data Usually layered on top of a lower-level physical storage medium Divided into logical units called files

High-Performance and Large-Capacity Storage: A Winning Combination for Future Data Centers. Phil Brace August 12, 2015

Automating Information Lifecycle Management with

Introduction to MapReduce

The Datacentered Future Greg Huff CTO, LSI Corporation

The Google File System. Alexandru Costan

Transcription:

Kinetic Open Storage Platform: Enabling Break-through Economics in Scale-out Object Storage PRESENTATION TITLE GOES HERE Ali Fenn & James Hughes Seagate Technology

2020: 7.3 Zettabytes 56% of total = in the cloud 90% Unstructured data 6.5 Zettabytes of unstructured data stored in the cloud 2010: 100 Exabytes 2

How big is 6.5 Zettabytes? 1 inch If you stacked 4TB hard drives side by side, they would circle the earth 3

6.5 Zettabytes Using today s architecture, it could cost cloud data centers: >$240,000,000,000/ye ar in CAPEX + 1 Year OPEX 4

The Opportunity for Change How do we get there? Ecosystem = Open and Software- Defined HDFS CEP H Where do we go from here? 5

Look at Legacy Opportunity Server Storage Application File System DB POSIX File System Volume Manager Driver RAID 1986 POSIX 1988 NTFS 1993 1990s to 2000s XFS 1993 Storage Server RAID Battery Backed RAM CACHE FC SAS Devices SAS Interface SMR, Mapping Cylinder, Head, Sector Drive HDA 6

Storage to Enable New World Use Cases Application File System DB Standard Device Recording POSIX File System Volume Manager Driver Storage Server RAID Battery Backed RAM CACHE F C Devices SAS Interface SMR, Mapping Cylinder, Head, Sector Drive HDA SAS Key/Value Interface Ethernet Connectivity 7

The Kinetic Open Storage Platform Open Source Key/Value API and libraries Open Source Interface Specification Object storage software partners Systems partners Storage now fully disaggregated from compute 8

Simplifying Storage Advancements 4K Sector Transitions = Greater Agility Shingled Magnetic Recording Advanced Management Data Security 9

Performance Opportunities Performance Raw throughput IO utilization Data streams to drive as written Drive handles space mgmt. (no file system metadata) 10

TCO Impact + + + Deploying a Kinetic based architecture could deliver: Up to 50% lower TCO 11

Single Drive 1 Additional Chip to Start 12

Libraries, API enable Applications Application Clustering Management Proprietary to System Vendor C++, Java, Python, Erlang, DIY Interconnect ProtoBuf TCP/IP/GbE GPL Standard Storage Proprietary to Seagate 13

Multiple Masters Application Clustering Management Proprietary to System Vendor Interconnect ProtoBuf TCP/IP/GbE GPL Standard Storage Proprietary to Seagate 14

Multiple Drives, P2P Operations Application Clustering Management Proprietary to System Vendor Interconnect ProtoBuf TCP/IP/GbE GPL Standard Storage Proprietary to Seagate 15

Goals of the Kinetic API Data movement Get/put/delete/getnext/getprevious Versioned (== for success), options Multiple masters Authentication/Integrity/Authorization Cluster-able Simple cluster configuration version enforcement 3rd party copy Management 16

Management (System Vendor) Configures the drive Network Authorized clients Monitors Health Statistics Logs Initiates recovery Change cluster version 3rdPartyCopy 17 17

Standard HDD Form Factor Connector re-pinned Two Ethernet connections Connector 18

System Implications No new ports Ethernet v. SAS switch 19

Network Implications No impact to Data Center Networking Traditional Architecture Kinetic Architecture 20

Performance Opportunities 21

Map of Operations 22

Performance Expectations Same normal performance expectations Sequential Write: 50 MB/s Random Write: 50 MB/s Sequential Read: 50 MB/s Random Read: 1.2x slower than traditional drives 23

Write Performance Results [PRELIMINARY] 24

Swift - Traditional 2 25

Swift - Kinetic 2 26

Swift - Kinetic 2 27

Basho Riak 2 28

Basho Riak - Kinetic 2 29

HDFS - Kinetic 30 30

Scality Kinetic Model Direct data path from clients to kinetic drives Native support for file, object and block Geo distribution across multiple sites Mix of replication and erasure coding Geo distributed metadata cluster 31

Kinetic Fits All Scale-Out Storage Object Storage Cloud Storage, Cloud backup, Cloud Archive / Cold Storage (Open Stack Swift, S3, Riak CS) Distributed File System Architectures Hadoop Distributed File System (HDFS), Google File System (GFS), Ceph, Windows Distributed File System (DFS), FhGFS, GlusterFS, Lustre Distributed Database and Memory Systems No SQL: Cassandra, Voldemort, Riak Memory: Memcached 32

Summary The Kinetic Open Storage Platform: Lowers TCO Disaggregates storage from compute Improves performance Increases innovation agility and efficiency More info at: developers.seagate.com 33