P R I S E A T S C A L E ► DDN now 1000 people worldwide! • Organic growth, R&D Investment • Acquisition ► Strategic Directions into AI, Multicloud, Exascale, Data Management ► Continued Commitment to AI and HPC marketplace • Strongest Portfolio for Data at Scale ► New Opportunities in Enterprise ► Customers are demanding more storage solutions from fewer suppliers ► Many start ups in the market and everyone is the fastest, but immaturity is risky in storage at scale ► No companies like DDN engage as a dedicated partner for the toughest data challenges.
P R I S E A T S C A L E Why DDN, Why Now? Emergence of AI Applications At-scale – Intersection of Enterprise and HPC AIOps Driven by Intelligent Infrastructure
P R I S E A T S C A L E YOUR STRATEGIC PARTNER FOR MODERN DATA WORKLOADS Virtualized Enterprise Software Defined Unified Storage High Performance Unified Storage E N T E R P R I S E AI Data Management HPC & Big Data Storage A T S C A L E
P R I S E A T S C A L E VMs Containers Databases & Analytics Objects GPU CPU VMWare, HyperV, NAS, VDI, VMWare, Openstack, SAS GRID, ANALYTICS, HPC, AI AI, Big Data & High-Performance Computing Our Expanded Ecosystem for End to End AI
P R I S E A T S C A L E At Scale Data Services for AI, Big Data and HPC Optimized AI Platforms For Every Use Case DDN BY High Performance Flash & Hybrid Unified Storage SW Defined Storage for Telco 5G, IoT and the Enterprise Simplicity and Control for Virtualized Environments Unlimited Performance for your most Demanding Workloads
Data Centric Enterprise ► As data volumes, complexity, diversity and value increase, our customers look for new ways to simplify operations and extract that value ► Our new combined IP will deliver on the promise of intelligent multicloud analytics that: • Simplifies, Predict and Automates • Puts the Right Data in the Right Place at the Right Time • Exposes deeper Value from All Data ► Beyond Storage: Unified analytics resolves broader business challenges across the entire ecosystem – On premises and multicloud
Storage market share ► Well over 10 EXAbytes of shipped capacity ► 150+ patents in data management ► “Top 5 AI vendors to watch” for choice and end to end ► Only company with > 16% of revenues invested back into AI and HPC R&D ► Fastest and Safest choice for Storage at Scale A T S C A L E >150 PATENTS #1 HPC Storage Vendor DDN >10 EXABYTES DELIVERED >7000 CUSTOMERS Intersect360 Research, 2019
with our partners and customers to build cutting edge solutions ► Tackling I/O optimization across HPC and Commercial ► Production-grade tools for I/O optimization (e.g. kubernetes, kafka) for autonomous ► Accelerating processing of Seismic data with Extreme- Scale Parallel I/O for Seismic Workflows ► Better efficiency and productivity of numerical weather and climate simulation A T S C A L E Advanced storage Monitoring Energy Oriented Center of Excellence I/O Optimization Energy Research Weather Modelling I/O Analytics AI and Big Data in HPC Research
Checkpoint Classify Manage Train Tune Infer User & Service Management Data Management Multi-Tenanted Security More Value from All your Data In One Place ► Your data in the right place at the right time ► Multicloud Ready ► The right user and service management to aid your environment’s efficiencies ► Comprehensive Security for Containers, Cloud and On Prem Transparent Tiering NFS SMB HDFS S3
Checkpoint Classify Manage Train Tune Infer Security GPU Network Protocol AI Optimizations HPC Optimizations Security GPU Network Protocol AI Optimizations Concurrency I/O Pattern Network Protocol HPC Optimizations Optimized across the board for both AI and HPC workloads and environments ► Deep Optimizations for both AI and HPC delivers for the highest efficiencies and the right capabilities ► AI at Scale: Container Aware, Direct GPU Integration, MultiRail Networking, Accelerated Protocols for AI Frameworks ► HPC: Any thread count, the toughest IO patterns, Dynamic Data placement, Leanest Datapath
Checkpoint Classify Manage Train Tune Infer Security GPU Network Protocol AI Optimizations HPC Optimizations User & Service Management Data Management Multi-Tenanted Security Security GPU Network Protocol AI Optimizations Concurrency I/O Pattern Network Protocol HPC Optimizations The Intelligent, Optimized Environment for AI and HPC ► Deep Optimizations for both AI and HPC delivers for the highest efficiencies and the right capabilities ► Your data in the right place at the right time Transparent Tiering NFS CIFS HDFS S3
time monitoring Comprehensive At Scale Data Management IME and EXA5 ► Ground-Up developed EXAScaler NVMe platform that delivers the best throughput and IOPs per NVMe ► Tight Integration with the most scalable HDD Platform with EXA5 data management ► Fully Automated Advanced Data Tiering designed for AI and HPC Super Fast Distributed Rebuild NVMe Optimized Data Layout and Flash Management Comprehensive Systems Management Dial In Erasure Coding Dynamic Data Placement Optimizes continuously EXAScale Efficiency that copes with real world at scale challenges
Stratagem Services State of jobs kept here with some metadata around stratagem state and history Parallel Actions 2019 2020+ Stratagem Scanning Engine Parallel scanning Stratagem State Stratagem Policy Engine Hot Pools Auto Purge Search Multicloud & Enterprise Data Management EXA Scale Change logs REST and UI MGMT …
Checkpoint Classify Manage Train Tune Infer Security GPU Network Protocol AI Optimizations HPC Optimizations User & Service Management Data Management Multi-Tenanted Security Security GPU Network Protocol AI Optimizations Concurrency I/O Pattern Network Protocol HPC Optimizations The Intelligent Multicloud always optimizing your data transfers to and from the cloud Multicloud cache Embargoed until March 23, 2020
Use Case Accelerate applications by achieving full GPU saturation on DGX Streamline concurrent and continuous deep learning workflows Flexible configuration with best technology and economics Seamless scaling to match evolving workflow needs Optimized for DGX platforms and NGC containers for DL and HPC Easy to deploy and manage with turnkey support from DDN and partners ®
SUPERPOD REFERENCE ARCHITECTURE ► AI400 + DGX-2 SuperPOD at-scale testing and validation published by NVIDIA: • The AI400 All-flash appliance delivers incredible sequential and random read performance, as required by the heaviest DL workloads. • Metadata performance scales well from 1 to 96 nodes, with no degradation as the number of nodes and threads increases. • The AI400 is a fully-integrated platform that’s easy to deploy. DDN provides excellent technical deployment and support services. ► RA document available from NVIDIA website NVIDIA DGX-2 SUPERPOD REFERENCE ARCHITECTURE
TO STORAGE ► Direct Memory Access from EXAScaler to GPU memory ► 2x-8x higher BW data transfers between Storage and GPU. ► 3.8x lower latency with no faulting and bounce buffers ► Stable and consistent latency as concurrency increases ► Lower consumption of host CPU or memory subsystem ► The GPU is the computing element with the highest IO bandwidth, e.g. 215 GB/s vs. the CPU’s 50 GB/s. ► Very fast access to petabytes of remote storage faster than even the page cache in CPU memory. PCIe SW CPU RAM GPU Network Bounce Buffer PCIe GPUDirect PCIe SW CPU RAM GPU Network GPUDirect Today
claimed as the property of others. Any statements or representations around future events are subject to change. Thank You! Keep in touch with us. 9351 Deering Avenue Chatsworth, CA 91311 1.800.837.2298 1.818.700.4000 company/datadirect-networks @ddn_limitless [email protected]
YOUR DATA RAPIDLY, EASILY, AND WITH CONFIDENCE • Data migration for massive data sets • Move data, maintain permissions and ACLs • Migrate your data safely • Automated migration • Track migration through an intuitive GUI
for a Distributed at Scale World Group A Group C Group B AUDIT ► Rogue clients, VMs and containers cannot gain root on filesystem ► Secure Shared Key or Kerberized Auth. of Clients/Containers ► Multi-Tenancy exports only specific datasets to users ► Secure Audit Logging ► Data at Rest Encryption for Drives and NVMe* Encryption *NVMe SED available Q1 2020 End to end Data Protection
and HPC ► Containers bring security challenges adding higher chance of malicious root access from clients ► Containers multiply the challenge of managing concurrency and load balancing across systems ► Capability class GPU systems such as DGX require fine grained managed of IO path across multiple networks. *NVMe SED available Q1 2020 Compromised containers get restricted privileges Parallel Management of Mounts across Network Path Simple Deployment. Completely Native Container load Balancing Multirail gets higher single system performance
itpresstour
4su_para
tohutohu
oracle4engineer
spycwolf
cmusudakeisuke
hogelog
uhyo
nobutomurata
asei
.jpeg){kind=avatar}
arthur1
kaminashi
aki_iinuma
orderedlist
thekraken
chriscoyier
eileencodes
shlominoach
philhawksworth
malarkey
garrettdimon
aarron
geoffreycrofte
tammyeverts
