©2012 DataDirect Networks. All Rights Reserved. Accelerates High Performance Computing and Big Data Application with Breaking through I/O Bottleneck 2012/10/26 Philip Zhu Director, North Asia, HPC
©2012 DataDirect Networks. All Rights Reserved.
Accelerates High Performance
Computing and Big Data Application
with Breaking through I/O Bottleneck
2012/10/26
Philip Zhu
Director, North Asia, HPC
©2012 DataDirect Networks. All Rights Reserved.
DDN | We Accelerate Information Insight
2
DDN is a Leader in Massively Scalable Platforms and
Solutions for Big Data and Cloud Applications
► Established: 1998
► Main Office: Sunnyvale, California, USA
► Employees: 600+ Worldwide
► Worldwide Presence: 16 Countries
► Installed Base: 1,000+ End Customers; 50+ Countries
► Go To Market: Global Partners, Resellers, Direct
World-Renowned & Award-Winning
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
DDN | Massive Scalability Focus
3
DDN powers
organizations to
accelerate discovery
to find new insights for
drug, energy and
scientific discovery.
The world’s largest
and most critical HPC
systems depend on
DDN.
DDN brings
cloud-enabled business
& social networkers
together – allowing
them to share and
collaborate on data.
The world’s largest web
and cloud infrastructures
depend on DDN.
DDN empowers
data-driven
organizations to make
informed, real-time
decisions to protect
society.
The world’s most
secure governments
and organizations
depend on DDN.
DDN provides the tools
for creative organizations
to create and broadcast
the richest possible
content.
The world’s most creative
and mission critical
content organizations
depend on DDN.
Media Web/Cloud HPC Security
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
DDN | Massive Scalability Focus
4
Media Web/Cloud HPC
Life Science
Financial Service
Energy Exploration
Manufacturing
Big Data Analytics
Web 2.0
Social Networking
Service Providers
Cloud Computing
Intelligence
Defense
Surveillance
Homeland Security
Broadcasting
Video Production
Film Animation
Video On Demand
Security
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
The Broadest Big Data Portfolio
5
Parallel File Storage SAN File Storage Enterprise NAS
EXAScaler™
10Ks of Clients
1TB/s+
GRIDScaler™
1Ks of Clients
1TB/s+
Scale-Out NAS
xSTREAMScaler™
100s SAN/LAN Clients
HSM Capable
Streaming Optimized
NAS Scaler™
1-16 NAS Servers
Fully-Featured
High Performance
Silicon Storage Architecture Storage Fusion Architecture
WOS®
256 Billion Objects
GeoReplicated
Cloud Foundation
Mobile Cloud Access 6GB/s in Real-Time
1,200 Drives: 2 Racks
2GB/s, 350K IOPS
120 Drives in 8U
15GB/s, 1.2M IOPS
1,200 Drives: 2 Racks
Embedded Computing
40GB/s/1.7M IOPS
1,680 Drives: 2 Racks
Embedded Computing
Platforms Support Any Drive Technology
to Optimize Data Intense Workflows
SAS SATA SSD
Object Storage
9900 6620 10K 12K
File Storage
Storage Engines
DirectMon
Big Data Management Made Easy
Infrastructure Management Media
DDN Confidential
©2012 DataDirect Networks. All Rights Reserved.
DDN | Storage Fusion Architecture (SFA)
6
Accelerating Big Data and Cloud, Optimizing TCO Over 1 Million Lines of S/W Code – First Customer Shipped 2008
Designed Specifically for Big Data and Cloud Workloads
Parallel State-Machine Design
Maximum Performance, Lowest
Latency
Virtualized Processing
Optimized Environment for Big Data
Application Hosting
Robust Data Protection
Quality of Service and Performance
Without Compromise
Flexible & Massively Scalable
Best-In-Class Scalability and Density
Storage Fusion Architecture™
[Core Storage S/W Engine]
In-Storage Processing™ Engine & DMA Driver
Dir
ec
tMo
n™
: In
fra
str
uc
ture
Ma
na
gem
en
t
‘Scaler File System Family
Low-Latency Connect: FC, IB, Memory
Interrupt-Free Storage Processing
ReACT™ Adaptive Cache Technology
DirectProtect™ Data Integrity Management
Quality of Service Engine
Storage Fusion Fabric™
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
Smarter Big Data Analytics
Infrastructure is A Reality
7 6/8/12
Volume Velocity Variety + +
Manage More Data
Than Traditional
Systems
Accelerate Runtimes
With The Highest
Performance AND
Performance
Efficiency
Best-In-Class Mixed
Mode Performance &
Optimized for The
Right Media For Your
Analytics
Eliminate Bottlenecks & Master:
©2012 DataDirect Networks. All Rights Reserved.
DirectMon™
8
DDN | DirectMon™ (new for Q312)
Storage Management Made Simple
► A powerful, intuitive single pane of
glass to monitor & manage your
environment
► Simplify the administration of DDN
SFA and ‘Scaler* environments
► Leverages DDN SFA 1.5 API
*GRIDScaler Support Immediately
EXAScaler and Hadoop in 2H12.
Site Overview
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
DDN | World-Leading Deployments
9
HPC Cloud & Web
Infrastructure
Professional
Media
Security
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
DDN | The “Big” In Big Data
98% Fortune 500 ► Customers of
Online File Service
► PBs of Cloud
Storage
768 Thousand ► # of Supercomputer
CPUs
► 35PB Storage
Processing System
23 Million ► Broadcast Viewers
► 9PB of Real-Time
Play-To-Air
Systems
If any company is well poised to take on the challenges of exascale
computing and big data, it's DDN, since this is its heritage. 451 Group
10 6/8/12
©2012 DataDirect Networks. All Rights Reserved.
DDN | Solving Big Data Challenges
11
“We are very pleased to work with DDN on a
solution that reduces the complexity of our
storage environment and accelerates our
research efforts.”
“We are very happy with the performance & the
flexible way we store the data, and scale. We
are talking 1-2 racks as compared to 6-7 racks.”
“DDN works all the time – there’s never a
need to take it offline, reformat it, or fine-tune
it. In our business there is no such thing as
just good enough.”
SFA technology is used by for massive internet
content analysis. To support this project, DDN
partners with Vertica to drive mission-critical,
real-time insight.
Worldwide Gene Sequencing Leader
Worldwide Cloud Storage Leader
Largest East-Coast Post Studio
US Government Intelligence Agency
6/8/12
©2012 DataDirect Networks. All Rights Reserved.
SFA12K™ | Models
™
Maximum Drives
System Interface
Drive Types
System Capacity
Bandwidth
Cache IOPS
Flash IOPS
In-Storage
Processing™
1,6801 1,6801 1,680
FDR IB
16Gb FC2
FDR IB
16Gb FC
FDR IB
10/40GbE
3.5” & 2.5” SSD, SAS & SATA (inter-mixable)
6.72PB (w/ 4TB HDDs)1
20GB/s
(raw I/O)
40GB/s
(raw I/O)
20GB/s
(file I/O)
850K 1.7M 850K
700K 1.4M 700K
N/A N/A Yes.
ExaScaler, GridScaler
Customer Provided
SFA12K-20 SFA12K-20E SFA12K-40
©2012 DataDirect Networks. All Rights Reserved.
SFA12K-40 Benchmark
13 Confidential Information 6/8/12
©2012 DataDirect Networks. All Rights Reserved.
GRIDScaler P
ara
llel F
ile S
tora
ge
Ap
plia
nce
Massively Scalable Parallel File Storage Appliance
► Easy to deploy, All-in-One appliance
based on IBM GPFS technology
► Scalable building block architecture
• 200GB/sec+ and 100,000s of IOPS
► Feature-Rich, Enterprise Grade
Quality and High Availability with no
single point of failure
► DirectMon centralized configuration
and monitoring solution
©2012 DataDirect Networks. All Rights Reserved.
12KE GRID is a Hit!
- Argonne National Labs - Mira
► Argonne – Mira • 48 racks of BGQ @ 1024 nodes per rack
o 1.6 GHz 16-way core processor and 16 GB RAM per node
• 384 I/O nodes
• Four Mellanox IS5600 QDR switches (GPFS Infiniband Verbs = Main Storage I/O protocol)
• For a total of 768K cores, 768 terabytes of RAM, and a peak performance of 10 petaflops.”
► Scratch file system (SFS) – 240+ GB/s • 16 SFA 12K-20E couplets
• Each w/ 10 SS7000 enclosures, 560 3TB SATA drives, 32 200GB SSD’s
► Home file system (HFS) • 3 SFA 12K-20E couplets, each w/ 5 enclosures, 240 3TB SATA drives, 32 200GB SSD's
• Embedded vm's are dedicated GPFS NSD servers, with GPFS sold and supported by IBM.
► Test and Development filesystem (T&D) • 1 SFA 12K-20E couplet, w/ 10 SS7000 enclosures, 560 3TB SATA drives, 32 200GB SSD's
• Embedded vm's are dedicated GPFS NSD servers, with GPFS sold and supported by IBM.
► HPSS Cache • 1 SFA 12K-40 couplet, w/ 10 SS7000 enclosures, 560 3TB SATA drives, 32 200GB SSD's
©2012 DataDirect Networks. All Rights Reserved.
Integration with Web Object Scaler
► Built for collaboration
across geos
► Simulate on GRIDscaler
and distribute using
WOS
► Ingest using WOS
access (NFS and CIFS)
and simulate on
GRIDscaler
► Back up files safely to
the WOS cloud for
disaster recovery
CIFS
Access
Clustered
NFS Access
Simulations
using Parallel File
Systems
©2012 DataDirect Networks. All Rights Reserved.
Reference Customers with GridScaler
Customers that require tremendous client-side bandwidth and/or would like to scale to 1000’s of client nodes.
Customers that require low latency communication for client access with high availability.
Customers that need to scale to multi-petabytes with simplified data management.
Customers that require the world’s leading density and energy efficient file storage system
©2012 DataDirect Networks. All Rights Reserved.
ExaScaler
• The World's Fastest Parallel File System at 300 GBs/sec
• Completely Open Source Technology
• Fault-Tolerant architecture to support Mission Critical Operations
• Superior Data and Performance Protection with a both S2A and SFA technologies
• Massive Scalability >10PB Volumes
• Allows concurrent file and directory access
• SFA10000E Features ExaScaler Natively Embedded
• Leading Rack, Energy and Disk Performance Efficiency
• Deployed by the HPC Storage Experts
7 Years of HPC
Leadership in
©2012 DataDirect Networks. All Rights Reserved.
Worldwide Performance Leader
DDN Exascaler Technology
Powers the World's
Fastest File Systems: • 300GB/s at CEA
• 250GB/s as US DoD
• 250GB/s at ORNL
• The World's Fastest Parallel File System
• RDMA Capable Single Client throughput of
2.5GB/s+
• Ideal for 10,000s of concurrent
client access to storage
Supports full single file concurrency
Supports full single directory
concurrency
• Network Flexible
Writes Natively to IP / IB
• S2A Real-time storage technology provides
guaranteed performance
Writes are as fast as Reads
No degradation in performance due to drive
errors, rebuild events or enclosure failures
©2012 DataDirect Networks. All Rights Reserved.
Designed for Scale
► Scalable Volume and Namespace • Scale from a few TBs to greater that 10 PB Volumes
► Scalable Bandwidth • From a few MBs to greater that 300 GB/sec of bandwidth
• Excellent price/performance
► Scalable Connectivity • Seamlessly scale from a single client to 20,000 clients concurrently
accessing storage
• Stripe files and directories across multiple OSSs
► Scalable Resiliency • No Single point of failure
• Add multiple paths to storage with additional data servers
► Scalable Simplicity • ExaScaler monitoring and event notification system
• Management console with SNMP trap display utility, logging, remote
administration and notification utility
IB with RDMA/
1GbE/10GbE
Add 1,000s of ExaScaler Clients
Add ExaScaler Building Blocks
©2012 DataDirect Networks. All Rights Reserved.
Sample Customers with ExaScaler
And many more…
©2012 DataDirect Networks. All Rights Reserved.
Case Studies
©2012 DataDirect Networks. All Rights Reserved.
Oakridge National Labs – Jaguar Super Computer
The HPC storage leader, DDN maintains
advantage through hyper-scale expertise
At 240GB/s – DDN has delivered the world's
most scalable bio-informatics storage solution
to ORNL
ORNL has over 400GB/s of DDN performance &
capacity efficient solutions
Today's supercomputers are tomorrow's
workgroup clusters, only DDN technology
scales up and down to meet and grow with
the complex requirements of life sciences
Gating Mechanism of Membrane Proteins
PI: Benoit Roux, Argonne National Laboratory/University of Chicago
Physical Basis of Recalcitrance to Hydrolysis of Lignocellulosic Biomass
Principal Investigator: Jeremy Smith, ORNL
Enabling
Petascale
Research
©2012 DataDirect Networks. All Rights Reserved.
When Total decided they needed to increase their seismic computation by
10X – they turned to DDN's ExaScaler File Storage technology.
With DDN technology – Total can scale 10X or 100X and is now ready for
whatever future requirements that may arise.
The world’s 10th largest company depends on
DDN Solutions to scale production operations…
Scalability Case Study: Seismic Workflow
© 2010 DataDirect Networks. All rights reserved.
©2012 DataDirect Networks. All Rights Reserved.
Thank You.
25 6/8/12