SMR in Linux Systems. Seagate's Contribution to Legacy File Systems. Adrian Palmer, Drive Development Engineering

Similar documents
October 30-31, 2014 Paris

Shingled Magnetic Recording (SMR) Panel: Data Management Techniques Examined Tom Coughlin Coughlin Associates

Linux SMR Support Status

Scaling the Areal Density Mountain. Dave Anderson Seagate

Linux Kernel Support for Hybrid SMR Devices. Damien Le Moal, Director, System Software Group, Western Digital

An SMR-aware Append-only File System Chi-Young Ku Stephen P. Morgan Futurewei Technologies, Inc. Huawei R&D USA

Whither Hard Disk Archives? Dave Anderson Seagate Technology 6/2016

HGST Shingled Magnetic Recording + HelioSeal Technology

HiSMRfs a High Performance File System for Shingled Storage Array

Shingled Magnetic Recording + HelioSeal Technology. Achieving Unprecedented Storage Capacity Through Innovation

GearDB: A GC-free Key-Value Store on HM-SMR Drives with Gear Compaction

SMORE: A Cold Data Object Store for SMR Drives

Using SMR Drives with Smart Storage Stack-Based HBA and RAID Solutions

File system internals Tanenbaum, Chapter 4. COMP3231 Operating Systems

Storage Systems for Shingled Disks

Invest in New Technologies or Divest in Market Share

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

Red Hat Enterprise 7 Beta File Systems

Skylight A Window on Shingled Disk Operation. Abutalib Aghayev, Peter Desnoyers Northeastern University

Advanced Format in Legacy Infrastructures More Transparent than Disruptive

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

Overview and Current Topics in Solid State Storage

Operating Systems. Operating Systems Professor Sina Meraji U of T

Overview and Current Topics in Solid State Storage

Operating Systems. File Systems. Thomas Ropars.

ZEA, A Data Management Approach for SMR. Adam Manzanares

MODERN FILESYSTEM PERFORMANCE IN LOCAL MULTI-DISK STORAGE SPACE CONFIGURATION

LightNVM: The Linux Open-Channel SSD Subsystem Matias Bjørling (ITU, CNEX Labs), Javier González (CNEX Labs), Philippe Bonnet (ITU)

Novel Address Mappings for Shingled Write Disks

File System Case Studies. Jin-Soo Kim Computer Systems Laboratory Sungkyunkwan University

DDN s Vision for the Future of Lustre LUG2015 Robert Triendl

CSE 451: Operating Systems Spring Module 12 Secondary Storage

COS 318: Operating Systems. Storage Devices. Jaswinder Pal Singh Computer Science Department Princeton University

Filesystems Lecture 10. Credit: some slides by John Kubiatowicz and Anthony D. Joseph

Topics. Lecture 8: Magnetic Disks

File System Case Studies. Jin-Soo Kim Computer Systems Laboratory Sungkyunkwan University

The Btrfs Filesystem. Chris Mason

Hard Disk Drives. Nima Honarmand (Based on slides by Prof. Andrea Arpaci-Dusseau)

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

The Datacentered Future Greg Huff CTO, LSI Corporation

BIg data era calls for Petabyte storage systems with

Making Storage Smarter Jim Williams Martin K. Petersen

Storage Systems : Disks and SSDs. Manu Awasthi July 6 th 2018 Computer Architecture Summer School 2018

COS 318: Operating Systems. Storage Devices. Vivek Pai Computer Science Department Princeton University

Modular Drive Process System Design for Optimal Factory Efficiency September 9, 2010

Marty Czekalski President, SCSI Trade Association - Emerging Interface and Architecture Program Manager, Seagate Technology

CSE 153 Design of Operating Systems

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

PASIG Disk Trends. Oracle Storage Technology 101 Session. Philippe Deverchère EMEA Storage CTO. September 16, 2014

I/O & Storage. Jin-Soo Kim ( Computer Systems Laboratory Sungkyunkwan University

Monday, May 4, Discs RAID: Introduction Error detection and correction Error detection: Simple parity Error correction: Hamming Codes

Ext4-zcj: An evolved journal optimized for Drive-Managed Shingled Magnetic Recording Disks

Open-Channel SSDs Offer the Flexibility Required by Hyperscale Infrastructure Matias Bjørling CNEX Labs

Linux File Systems: Challenges and Futures Ric Wheeler Red Hat

Announcements. Persistence: Log-Structured FS (LFS)

Seagate Point of View Cloud and Data Center Trend

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

High Performance Solid State Storage Under Linux

STORAGE SYSTEMS. Operating Systems 2015 Spring by Euiseong Seo

COS 318: Operating Systems. Storage Devices. Kai Li Computer Science Department Princeton University

QuickSpecs. What's New. Models. HP SATA Hard Drives. Overview

Long-term Information Storage Must store large amounts of data Information stored must survive the termination of the process using it Multiple proces

File System Internals. Jo, Heeseung

SSD/Flash for Modern Databases. Peter Zaitsev, CEO, Percona November 1, 2014 Highload Moscow,Russia

CSE 333 Lecture 9 - storage

File system internals Tanenbaum, Chapter 4. COMP3231 Operating Systems

GFS: The Google File System

Solid State Drives (SSDs) Jin-Soo Kim Computer Systems Laboratory Sungkyunkwan University

Mass-Storage Structure

NAND Flash-based Storage. Jin-Soo Kim Computer Systems Laboratory Sungkyunkwan University

PERSISTENCE: FSCK, JOURNALING. Shivaram Venkataraman CS 537, Spring 2019

Open Source for OSD. Dan Messinger

Segmentation with Paging. Review. Segmentation with Page (MULTICS) Segmentation with Page (MULTICS) Segmentation with Page (MULTICS)

CrashMonkey: A Framework to Systematically Test File-System Crash Consistency. Ashlie Martinez Vijay Chidambaram University of Texas at Austin

I/O CANNOT BE IGNORED

January 28-29, 2014 San Jose

Chunling Wang, Dandan Wang, Yunpeng Chai, Chuanwen Wang and Diansen Sun Renmin University of China

QuickSpecs. What's New. Models. HP SATA Hard Drives. Overview. HP 6G SATA SmartDrive Carriers

Alternatives to Solaris Containers and ZFS for Linux on System z

QuickSpecs. HPE SAS Hard Drives. Overview. What's New

<Insert Picture Here> End-to-end Data Integrity for NFS

Introduction to Open-Channel Solid State Drives and What s Next!

Disk Scheduling COMPSCI 386

CS370 Operating Systems

UNIT 2 Data Center Environment

Chapter 11: Implementing File Systems

File. File System Implementation. File Metadata. File System Implementation. Direct Memory Access Cont. Hardware background: Direct Memory Access

u Covered: l Management of CPU & concurrency l Management of main memory & virtual memory u Currently --- Management of I/O devices

Cold Storage: The Road to Enterprise Ilya Kuznetsov YADRO

Wednesday, April 25, Discs RAID: Introduction Error detection and correction Error detection: Simple parity Error correction: Hamming Codes

File System Case Studies. Jin-Soo Kim Computer Systems Laboratory Sungkyunkwan University

Longhorn Large Sector Size Support. Anuraag Tiwari Program Manager Core File System

I/O CANNOT BE IGNORED

Storage Technologies - 3

Main Points. File layout Directory layout

Open-Channel SSDs Then. Now. And Beyond. Matias Bjørling, March 22, Copyright 2017 CNEX Labs

CS370: Operating Systems [Spring 2017] Dept. Of Computer Science, Colorado State University

Storage Speed and Human Behavior. PRESENTATION TITLE GOES HERE Eric Herzog CMO and Senior VP of Business Development Violin Memory

Storage Systems : Disks and SSDs. Manu Awasthi CASS 2018

Introduction to I/O and Disk Management

Transcription:

SMR in Linux Systems Seagate's Contribution to Legacy File Systems Adrian Palmer, Drive Development Engineering

SEAGATE combines DIFFERENT TECHNOLOGIES in new ways to SOLVE customer data storage CHALLENGES 2

Shingled Magnetic Recording (SMR) Areal density growth curve lowest cost/gb 10 TB/in^2 HAMR with SMR BPMR with SMR HDMR with BPMR 1 PMR SMR SMR (Shingle Magnetic Recording) HAMR (Heat Assisted Magnetic Recording) BPMR (Bit Patterned Magnetic Recording) 0.1 PMR (Perpendicular Magnetic Recording) 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Data adapted from ASTC Technology Roadmap (http://www.idema.org/?page_id=5868) 3

Shingled Magnetic Recording (SMR) Forward-write only: Radial AND Rotational Image: Wood, R.; Williams, Mason; Kavcic, A.; Miles, Jim, "The Feasibility of Magnetic Recording at 10 Terabits Per Square Inch on Conventional Media," Magnetics, IEEE Transactions on, vol.45, no.2, pp.917,923, Feb. 2009 4

SMR Drive Types Drive Managed (DM) Mimics Traditional drives Backwards compatible Direct Replacement for conventional drives in conventional apps Host Managed (HM) Not backwards compatible Host required to manage data ordering for performance mitigation Extensions required in ATA and SCSI command sets Host Aware (HA) Combination of DM and HM. Backwards compatible / Able to use extensions in ATA and SCSI

Drive Managed Compatibility for Today Host Aware Performance for Tomorrow Drive Managed Host Managed Host Aware No change: regular SD drive Requires new device & FS Regular SD drive FS benefits from knowledge of media layout Host Aware: Capacity gains like Drive Managed Performance like Conventional 6

SMR Can we avoid it? Benefits Provides continued growth in Areal Density. Enables lower cost/gb disc drives Base of new technologies HAMR Support Readiness ZBC/ZAC specifications are nearing completion T10/T13 committees work actively progressing Availability Millions of DM drives shipped! Seagate s 8TB Archive HDD v2 drive is SMR DM in production, HA forthcoming ZBC: Zoned Block Commands ZAC: Zoned device ATA Commands

ZAC/ZBC Standards Inspired by SMR. Applicable to any media Separates media into bounded zones Write Pointer Zones Sequential write only for Host Managed (restrictive) Sequential write preferred for Host Aware (permissive) New common ATA/SCSI commands REPORT ZONES RESET WRITE POINTER OPEN/CLOSE/FINISH ZONE Requires communication with FS beyond simple Read/Write 8

SMR Friendly File System Requirements Forward-write only CoW requirement Zone aware (ZAC/ZBC) Boundaries New Commands REPORT_ZONES RESET_WRITE_POINTER OPEN/CLOSE/FINISH ZONE New algorithms Defragmentation Goals Optimally Directed Writes efficient streams Excellent Reads streaming File defragmentation Metadata handling Backwards compatibility Provide reference design for other file systems 9

SMRFS -EXT4 Default FS of many distros Build upon strengths Popular FS Stability Market Acceptance Compatible on-disk format Popular in commercial storage applications Minimal architecture changes; only rearrangement of existing data 10

Proposed Stages EXT4 SMRFFS Project Scope Steps v7 Enforce Host Managed v5 v4 Garbage Collection Kernel Integration Utilities v6 1. cmd line arguments v1 2. internal handling changes v2 3. kernel stack changes v3 v2 Internal changes v1 v2 Specify mkfs Options for HA v2 EXT4 SCSI Disk v2 Kernel Stack v3 4. IOCTL integrations v4 5. Algorithm enhancements v5 6. Utility updates v6 7. Host Managed Compliance v7 12

Proposed Stack Changes V3 - Kernel Stack Changes, V4 Kernel Integration VFS VFS EXT4 Examine/Enforce I/O Ordering O_Direct IOCTL Page Cache I/O Scheduler SCSI S-A-T ATA AHCI Disk 13

State of project We ve done Laid out design, made prototypes Discussion at LSF Consensus of key developers We ve got to do A lot of work in a short time And we need community help! Ask how to contribute and how to get sample drives at our booth! Contact us after Vault at adrian.palmer@seagate.com or timothy.r.feldman@seagate.com Seagate Confidential 14

ext4 ext4 flex_bg ext4 dm-cow zfs btrfs ext4 ext4 flex_bg ext4 dm-cow zfs btrfs ext4 ext4 flex_bg ext4 dm-cow zfs btrfs ext4 ext4 flex_bg ext4 dm-cow zfs btrfs Throughput - MB/s File System Parameters Influence Performance For Drive Managed SMR 200 180 160 140 120 100 80 60 Drive Managed SMR - Performance by FS Parameters, for different Workloads drivea driveb drivec Mimic OS Load Mimic DVD Download 5meg files Deep Dir 1meg files Deep Dir Re-arranging file system parameters for CoW to enforce forward-writeonly improves performance of a DM-SMR-enabled system 15

Q&A Thank You! Attendees and Partners