Next Generation HPC Storage Initiative. Torben Kling Petersen, PhD Lead Field Architect - HPC



Similar documents
Lustre Networking BY PETER J. BRAAM

Seagate Lustre Update. Peter Bojanic

Quick Reference Selling Guide for Intel Lustre Solutions Overview

New Storage System Solutions

Sun Storage Perspective & Lustre Architecture. Dr. Peter Braam VP Sun Microsystems

Xyratex Update. Michael K. Connolly. Partner and Alliances Development

Easier - Faster - Better

Data management challenges in todays Healthcare and Life Sciences ecosystems

Globus and the Centralized Research Data Infrastructure at CU Boulder

Lessons learned from parallel file system operation

InfiniBand Software and Protocols Enable Seamless Off-the-shelf Applications Deployment

Secure Your Business with EVault Cloud-Connected Solutions

HPC Software Requirements to Support an HPC Cluster Supercomputer

Choices for implementing SMB 3 on non Windows Servers Dilip Naik HvNAS Pty Ltd Australians good at NAS protocols!

Seagate Cloud Systems & Solutions

Can High-Performance Interconnects Benefit Memcached and Hadoop?

An Alternative Storage Solution for MapReduce. Eric Lomascolo Director, Solutions Marketing

Inside Lustre HSM. An introduction to the newly HSM-enabled Lustre 2.5.x parallel file system. Torben Kling Petersen, PhD.

Percipient StorAGe for Exascale Data Centric Computing

POWER ALL GLOBAL FILE SYSTEM (PGFS)

Cray DVS: Data Virtualization Service

Achieving Mainframe-Class Performance on Intel Servers Using InfiniBand Building Blocks. An Oracle White Paper April 2003

Introduction to MPIO, MCS, Trunking, and LACP

IBM PureFlex System. The infrastructure system with integrated expertise

The Fastest Path to the Cloud Building Your SaaS Company on Force.com

EMC BACKUP MEETS BIG DATA

Flexible Scalable Hardware independent. Solutions for Long Term Archiving

Symantec Backup Appliances

Seagate HPC /Big Data Business Tech Talk. December 2014

Agenda. Enterprise Application Performance Factors. Current form of Enterprise Applications. Factors to Application Performance.

High Performance Computing OpenStack Options. September 22, 2015

Lunch and Learn: Modernize Your Data Protection Architecture with Multiple Tiers of Storage Session 17174, 12:30pm, Cedar

The Next Generation of Security Leaders

How To Create A Large Enterprise Cloud Storage System From A Large Server (Cisco Mds 9000) Family 2 (Cio) 2 (Mds) 2) (Cisa) 2-Year-Old (Cica) 2.5

InfiniBand in the Enterprise Data Center

Big data management with IBM General Parallel File System

From Ethernet Ubiquity to Ethernet Convergence: The Emergence of the Converged Network Interface Controller

MySQL Database Replication and Failover Clustering

Peter Duflo Managing Partner

Fusion Service Schedule Virtual Data Centre ( VDC ) Version FUS-VDC-7.1

Building Reliable, Scalable AR System Solutions. High-Availability. White Paper

EOFS Workshop Paris Sept, Lustre at exascale. Eric Barton. CTO Whamcloud, Inc Whamcloud, Inc.

Product Spotlight. A Look at the Future of Storage. Featuring SUSE Enterprise Storage. Where IT perceptions are reality

Agenda. HPC Software Stack. HPC Post-Processing Visualization. Case Study National Scientific Center. European HPC Benchmark Center Montpellier PSSC

Product Brief. DC-Protect. Content based backup and recovery solution. By DATACENTERTECHNOLOGIES

A Tour of the Linux OpenFabrics Stack

Oracle Maximum Availability Architecture with Exadata Database Machine. Morana Kobal Butković Principal Sales Consultant Oracle Hrvatska

Sun in HPC. Update for IDC HPC User Forum Tucson, AZ, Sept 2008

February, 2015 Bill Loewe

IBM Global Technology Services September NAS systems scale out to meet growing storage demand.

Network Function Virtualization Using Data Plane Developer s Kit

Red Hat Enterprise Linux 6. Stanislav Polášek ELOS Technologies

Quantum StorNext. Product Brief: Distributed LAN Client

SMB Direct for SQL Server and Private Cloud

Distributed File System Choices: Red Hat Storage, GFS2 & pnfs

Cisco Unified Computing Remote Management Services

SMB Advanced Networking for Fault Tolerance and Performance. Jose Barreto Principal Program Managers Microsoft Corporation

Intel Network Builders: Lanner and Intel Building the Best Network Security Platforms

Virtual Compute Appliance Frequently Asked Questions

Performance, Reliability, and Operational Issues for High Performance NAS Storage on Cray Platforms. Cray User Group Meeting June 2007

Organizations that are standardizing today are enjoying lower management costs, better uptime. INTRODUCTION

Hyper-V over SMB: Remote File Storage Support in Windows Server 2012 Hyper-V. Jose Barreto Principal Program Manager Microsoft Corporation

SAP HANA Operation Expert Summit BUILD - High Availability & Disaster Recovery

IBM Cloud Builder Professional Services. and Remote Managed Services

SEP Disaster Recovery and Backup Restore: Best

SCALABILITY AND AVAILABILITY

Why is it a better NFS server for Enterprise NAS?

Nine Considerations When Choosing a Managed Hosting Provider

Maximum Availability Architecture

Backup and Recovery FAQs

The Service Availability Forum Specification for High Availability Middleware

Vendor Update Intel 49 th IDC HPC User Forum. Mike Lafferty HPC Marketing Intel Americas Corp.

Network Attached Storage. Jinfeng Yang Oct/19/2015

PRIMERGY server-based High Performance Computing solutions

Tiered Adaptive Storage

Cisco Active Network Abstraction Gateway High Availability Solution

The Red Hat Enterprise Linux advantages over Oracle Linux

Inside The Lustre File System

Kriterien für ein PetaFlop System

WHITE PAPER Improving Storage Efficiencies with Data Deduplication and Compression

Intel DPDK Boosts Server Appliance Performance White Paper

Hyper-V over SMB Remote File Storage support in Windows Server 8 Hyper-V. Jose Barreto Principal Program Manager Microsoft Corporation

Running Native Lustre* Client inside Intel Xeon Phi coprocessor

Transcription:

Next Generation HPC Storage Initiative Torben Kling Petersen, PhD Lead Field Architect - HPC

2 Forward Looking Statement The following information contains, or may be deemed to contain, "forward-looking statements" (as defined in the U.S. Private Securities Litigation Reform Act of 1995) which reflect our current views with respect to future events and financial performance. We use words such as "anticipates," "believes," "plans," "expects," "future,"' "intends," "may," "will," "should," "estimates," "predicts," "potential," "continue" and similar expressions to identify these forward-looking statements. All forward-looking statements address matters that involve risks and uncertainties. Accordingly, you should not rely on forward-looking statements, as there are or will be important factors that could cause our actual results, as well as those of the markets we serve, levels of activity, performance, achievements and prospects to differ materially from the results predicted or implied by these forward-looking statements. These risks, uncertainties and other factors include, among others, those identified in "Risk Factors," "Management's Discussion and Analysis of Financial Condition and Results of Operations'' and elsewhere in the company s 20-F filed with the SEC. Xyratex Ltd undertakes no obligation to publicly update or review any forwardlooking statements, whether as a result of new information, future developments or otherwise.

3 Unique and Deep Understanding of Storage Market

4 Leading OEM Provider of Digital Storage Technology SI: Largest independent supplier of Disk Drive Capital Equipment ~ 50% of w/w disk drives are produced utilizing Xyratex Technology ~ 75% of w/w 3.5 LFF disk drives NSS: Largest OEM Disk Storage System Supplier 33% WW OEM Market Share in 2009, 5 Tier-1 OEM s 16% of worldwide external storage capacity shipped in 2009 (IDC) 2010 Revenue ($1,603M) $1,260 $343 > 3.0 Exabyte's of storage shipped in 2010 ~ 139,000 storage enclosures shipped in 2010 SI NSS

5 Xyratex: Global Presence Over 2150 Employees Worldwide

6 Global Customer Relationships Strategic relationships established with both market & technology leaders

Xyratex HPC & Lustre

Xyratex in HPC Via our OEM base, Xyratex has had a successful history in supplying storage building blocks for HPC storage solutions In 2010, Xyratex shipped over 3.0 Exabytes of external storage capacity worldwide, ~750 Petabytes (25%) was deployed into HPC environments Benefits for our HPC OEMs: 30+ years of storage technology development & innovation Proven track record of overcoming storage integration challenges Enables OEM Partners to focus on their core competencies HPC Compute Expertise Creates sustainable leadership & competitive advantages for their storage systems Improved business flexibility via a global logistics capability Xyratex is making strategic investments to strengthen our HPC market position Capabilities, Technology (Hardware & Software), Product Breadth & Scale 8

9 The ClusterStor Acquisition ClusterStor Clustered File System start-up, led by the founder of Lustre Dr.Peter Braam World-class development team, deep Lustre expertise Have 3 of top-5 Lustre developers in the world Skilled technical staff with over a 100 years of combined Lustre experience Xyratex Acquisition Dr. Braam and the ClusterStor team have joined Xyratex, where they will form a new File System development and support team Attracting other key Lustre talent Peter Bojanic, former head of Lustre at Sun/Oracle Makia Manich, Isaac Huang, Nathan Rutman and others Dr. Braam will lead this team as SVP of Storage Software ~150 people worldwide Dr. Peter Braam SVP of Software Xyratex

Strong Lustre Adoption Continues in Top500 Lustre s share of Top 100 HPC sites is actually growing Now 8 of Top 10 Systems in Nov 2010 Now 65 of Top 100, up from 47 in previous year Top 100 65 21 3 11 Was 47 Lustre GPFS PanFS Other Top 10 8 2 0 0% 20% 40% 60% 80% 100% 10

Lustre Community Xyratex s Position Oracles position on Lustre has changed Shut down all Lustre development efforts Xyratex is 100% committed to supporting Open Lustre in the Linux community and future roadmap Key part of Xyratex long term strategy, supported by significant and ongoing investment Industry leading Lustre architecture and development team led by Dr. Peter Braam and Peter Bojanic Focused of Lustre 2.x, significant investment in stability/testing/quality Development focused both on core Lustre file system features and application integration Opportunity for Xyratex partners to influence Xyratex Lustre release/features/roadmap Lustre community is responding with a number of key initiatives both in the US and Europe US HPCFS - (US, Europe and ASIA)- Founding member, Intel led OpenSFS (US)- in process held up by IP ownership issues, DOE funding 11 Europe European Open File System SCE (OFS)- Founding Member PRACE- partnering with CEA and Cines on Lustre based Exascale demonstration system

12 Xyratex Truly Differentiated in the Lustre Market Designs, Develops and Tests the World s Best Purpose Built Storage Platforms OEM Business Model Economies of Scale Number 1 OEM Storage Supplier in the world (IDC) Linux based Storage Appliance Middleware Development World-Class Clustered File System Development & Support Expertise Storage Cluster Management Framework

LNET Channel Bonding

14 LNET Closer to the metal Vendor Network Device Libraries Support for multiple network types including routing API Lustre Network Drivers (LNDs) Sun LNET Library Network I/O (NIO) API Lustre Request Processing Similar to Sandia Portals with some new features Move small and large buffers Use RDMA Generate events Zero-copy marshalling libraries Service framework and request dispatch Connection and address naming Generic recovery infrastructure Legend: API Not supplied by with Lustre software Not portable Portable Lustre component

Channel bonding Channel Bonding Why?? LNet is network module designed to exchange messages between Lustre nodes. Channel bounding is needed to extend network bandwidth between nodes, and to allow to send many messages parallel, each in own network channel. Channel Bonding will also allow for redundancy in addition to improved bandwidth Some channel bounding modes already exists in Linux System bounding some networks channels merged in single logical channel on system side and lustre see that as one channel with single address space. LND based bounding currently supported on socklnd (tcp network module). In that case nodes use extended HELO protocol to send all addresses from node to remote peer and peer a look to these addresses to send request in round robin order. None of these modes support a channels bonding between different network types such as IB and Ethernet networks. To avoid that problem we are developing an implementation of channel bonding inside the LNet module. 15

16 Other Xyratex enhancements to Lustre Open Source samba-xattr-fixes.patch - Improvements for xattr handling for Samba. quota-type-fix.patch - Allows Lustre 2.0 to be started on top of an 1.8 FS with quota enabled. obd-device-list.patch - Allows more then 8192 obd devices per host. mdt statistics - two patches related to speeding up mdt performance patches related to backup / restore 2.x FS. patches related to build process Specifically for ldiskfs / lustre / kernel requirements and build fixes. Several other projects (such as Channel Bonding) is currently underway and the roadmap will be presented at LUG2011 in Orlando (April 12-14 th ).

17 Torben_Kling_Petersen@xyratex.com