Bjorn Andersson HPC Advisory Workshop, March 21-23 2011 Lugano, Switzerland Who Is Bluearc

Storage Performance Through Standards

Bjorn Andersson HPC Advisory Workshop, March 21-23 2011 Lugano, Switzerland Who is BlueArc

Private Company, founded in 1998

Headquarters in San Jose, CA

Highest Performing NAS in the Industry

Proven in the most demanding HPC environments

6th Generation Product

Patented Hardware Accelerated Architecture

Doubled performance and scale with each platform release

. Atomic Weapons Establishment . Massachusetts Institute of Technology . Brookhaven National Laboratory . NASA . Baylor College of Medicine . National Cancer Institute . Cambridge University . National Heart, Lung, & Blood Institute . Chevron . National Oceanic and Atmospheric Administration . Columbia University . National Renewable Energy Laboratory . Cold Spring Harbor Laboratory . Oak Ridge National Labs . Commissariat a L’Energie Atomique . Ontario Institute for Cancer Research . Cray, Inc. . Penn State University . Duke University . Princeton University . European Bioinformatics Institute . Purdue University . Fermi National Accelerator Laboratory . Renaissance Computing Institute . Fred Hutchinson Cancer Research Center . RWTH Aachen . Genentech . Sandia National Laboratories . Georgia Institute of Technology . Sanger Institute . HECToR Partners . Stanford University . Idaho National Laboratory . Tokyo Tech . Jet Propulsion Laboratory . University of California Los Angeles . Johns Hopkins University . University of Michigan . Lawrence Berkeley National Laboratory . University of Minnesota . Lawrence Livermore National Laboratory . Vanderbilt University . Merck & Co. . Washington University in St. Louis 3 © 2010 BlueArc Corp. Proprietary and Confidential Trend: The Rise of Unstructured Data

Worldwide File and Block Disk Storage Systems, 2005-2011 50000 45000 40000 35000

30000

25000 (PB) 20000 15000 10000 5000 IDC Storage 0 Shipment Forecast 2006 2007 2008 2009 2010 2011 2012 2006 to 2012 Block Based File Based Source: IDC, 2007

By 2012, over 80% of total disk storage system capacity shipped will be to support file-level data

- IDC: Worldwide Disk Storage Systems 2008.2012 Forecast

Build for multi-petabyte scale and HPC performance requirements while using standards and adding features more typically found in a general purpose file systems

GPFS Lustre ZFS FAT32 HFS+ BFFS HPFS NAS:UFS PFS ISO 9660 ext2 AFS XFS CXFS CIFS NFS (standard)EFS NTFS FFS GFS CacheFS JFS VxFS (… andNFS CIFS)DFS QFS SMB NSS hfs

Standard clients User Benefits: Choice with Multiple, Simplicity and familiarity Competing Implementations

Optimized Robust, Bigger pipes Operation Scalable At Scale Performance Hardware Acceleration (bandwidth&iops) High Availability Millions of files Easy Capacity Advanced Tiered Architecture, Scaling Open backend

No compromise design: • Offload engine . Parallel processing of core file system for high speed (FPGA offload engine) data transfer . Simultaneous access to CIFS, NFS and iSCSI (pNFS mid 2011)

• CPUs for . CPU runs data management functions unburdened data not already burdened by file system management . Object File System Architecture for Intelligent, flexible data management (SiliconFS) File . Storage virtualization and thin • Intelligent file Metadata provisioning Index Map objects for high efficiency User data

Server Running Dedicated Routing Dedicated Routing Appliance Routing Software Appliance with Custom OS with Routing in Hardware

General PerformanceSoftwareLimit Software Hardware Purpose Appliance Appliance

Server

Server Running Dedicated File Server Dedicated File Server

CIFS/NFS Software Appliance with Custom OS PerformanceSoftwareLimit Appliance with File System in Hardware

. As concurrent connections or # of Line Rate Speed clients scale – Performance level increases

linearly – No fall off due to CPU utilization – Much higher level of maximum performance – If pushed to maximum sustain 100% indefinitely BlueArc AdvantageBlueArc . Benefits Software Solution

Performance – More users per filer – More functions per filer – Fewer filers & licenses Concurrent connections or # of clients – Simplified management Scaling Storage or Turning on Features

Robust NFS Performance

Scales

Independent ofsizes Independentblock Independent of Read/Write mix of Read/Write Independent

Source: www.spec.org

Mercury 55 Mercury 110 Titan 3100 Titan 3200

Product Class Lower Mid-range Mid-range Mid-range High End

Cluster Nodes 2 Up to 4 Up to 8 Up to 8

Max Storage Capacity 4 PB 8 PB 8 PB 16 PB

Performance (specSFS IOPS) 60,000 100,000 100,000 200,000

NFS Throughput 700 MB/s 1100 MB/s 1200 MB/s 1900 MB/s

Storage Options All BlueArc storage array options are available with each platform

Software / File Services All software and filesystem options (NFS, CIFS, iSCSI) available

Users Network Storage Cluster Tier 3 Back-end Storage SAN • Deduplication • Encryption • Compression • Existing NAS Tier 0 • Solid State Cache

SAN Tier 1 • High Performance

Tier 2 • High Capacity

The seamless migration of . Automatic and transparent data migration between tiers data across storage tiers within a single namespace . Rules-based policy engine reduces manual intervention . Third-party or external storage devices as an integrated tier Ease data management and reduce costs . Reduced dependence on high performance tier for peak demands

Most Competitors BlueArc

. External or internal Direct . External shared switched Attached Storage Fabric SAN . 2-way or N-way clustering . N-way clustering . No automated storage . Multiple storage tiers tiering

In-house codes

Applications Adds up to

random and impossible to predict workload ● ● Instruments ● Researchers Back to the Traditional Network File System

Lots of Goodness BUT • Proven architecture • Enterprise features • No Throughput Aggregation • Open, standard protocols Beyond Line Rate • Open storage philosophy

. pNFS adds parallel I/O to the NFS protocol – Eliminates the file server bottleneck – Provides parallel data paths, even for a single file . pNFS is part of the NFSv4.1 standard – Approved by IETF Dec, 2008 – RFCs completed editorial review Oct, 2009 – RFC numbers issued Jan, 2010 . Multiple implementations are in development – Client software is expected to be embedded in leading OS distributions . The Only Industry Standard Parallel File System

GPFS Lustre ZFS FAT32 HFS+ BFFS HPFS ParallelUFS FS: PFS ISO 9660 ext2 AFS XFS CXFS CIFS LustreEFS, GPFS NTFS FFS GFS CacheFS JFS VxFSpNFS (standard)NFS DFS QFS SMB NSS hfs

Environments

Shared Repositories

Virtualization ScalingUp

Home Directories Cloud Storage

Compute Cluster Temporary Workspace Streaming (Scratch)