Upload
insidehpc
View
225
Download
0
Embed Size (px)
DESCRIPTION
In this deckt, Uday Mohan from DataDirect Networks presents: DDN GS7K - Easy-to-deploy, High Performance Scale-Out Parallel File System Appliance. High performance computing is critical in commercial markets, spanning a wide range of applications across multiple industries, and this trend is only growing. The GS7K from DDN will help bring the latest high-performance storage technologies to more of these markets, connecting companies to their next innovations faster while satisfying their enterprise standards.” Watch the video presentation: http://wp.me/p3RLHQ-d99
Citation preview
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
GS7K - Easy-to-deploy, High Performance Scale-Out Parallel File System Appliance
September 2014
Under Embargo till Oct 1, 12:30 a.m. ET
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
DDN | Who We Are
► Markets: Enterprise Big Data, Cloud, HPC
► Solutions: Platforms, File Systems, Object Storage
► Customers: 1,000+ in 50 Countries
► Go-To-Market: Direct, Partner Assist
► Employees: 550 in 20 Countries
► Headquarters: Santa Clara, CA
► History: Founded in 1998 and Profitable
We Solve Big Data Lifecycle Management Challenges at Large Scale
• The World’s Largest Private Storage Company
• NDA Confidential
2
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Big Data & Cloud Infrastructure DDN’s Award-Winning Product Portfolio3
Analytics Reference Architectures
EXAScaler™
10Ks of Clients1TB/s+, HSM
Linux HPC ClientsNFS & CIFS [2014]
Petascale Lustre® Storage
Enterprise Scale-Out File Storage
GRIDScaler™
~10K Clients1TB/s+, HSM
Linux/Windows HPC ClientsNFS & CIFS
SFA™12KX48GB/s, 1.7M IOPS1,680 Drives in 2 RacksOptional Embedded Computing
SFA770012.5GB/s, 450K IOPS60 Drives in 4U228 Drives in 12U
Storage Fusion Architecture™ Core Storage Platforms
SATA SSD
Flexible Drive Configuration
SAS
SFX™ Automated Flash Caching
WOS® 3.032 Trillion Unique Objects
Geo-Replicated Cloud Storage256 Million Objects/Second
Self-Healing CloudParallel Boolean Search
Cloud Foundation
Big Data PlatformManagement
DirectMon™
CloudTiering
Infinite Memory Engine™ [Tech Preview]Distributed File System Buffer Cache
WOS700060 Drives in 4U
Self-Contained Servers
Adaptive Transparent Flash Cache SFX API Gives Users Control [pre-staging, alignment, by-pass]
ddn.com©2013 DataDirect Networks. All Rights Reserved.
Scale Out | NAS vs Parallel File Storage
Scale-Out NAS Does Not Mean Parallel!
NAS - Point:Point TechnologyClients Can Only See One Server At A Time Locking Engines Not Designed For Massive Parallelism
Parallel File Systems – Removes NAS bottlenecksRead/Write From Many Servers in Parallel – ConcurrentOrder of Magnitude Higher Performance than NAS
Object-Based Locking; No Dedicated Storage Network; Scalable To 100+ PBsObject-Based Locking; No Dedicated Storage Network; Scalable To 100+ PBs
Support for RDMA Communication & Large PacketsMakes Best Use of High Performance Networks!
Big Data Cluster
Parallel Storage Cluster
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
The News: Introducing GS7K
► DDN is leveraging it’s HPC expertise and 3 generations of Embedded technology to bring to market the GS7K – an All-in-one, Scale-out Parallel File System Appliance.
► Combines DDN’s Storage Fusion Architecture with the scalability of IBM’s GPFS file system to introduce the industry’s first scale-out parallel file system appliance complete with enterprise-class features, NAS access and Cloud tiering capabilities
► Customers no longer have to trade off between performance and enterprise features: they can have the best of both worlds including massive performance and scalability in a scale-out parallel file system that also offers the simplicity and rich feature set in an all-in-one appliance
► Designed to help customers in data intensive industries like oil and gas, FSI, and life sciences jumpstart their first Big Data projects with an optimally sized and priced enterprise appliance
► Engineered for performance, GS7K can deliver ~12 GB/s per scale-out building block, making the appliance the most powerful and scalable appliance of its size
► Offers a building block approach so customers can start small with 60 drives in 4U and expand capacity by adding up to 4 additional 4U 84 drive enclosures. Multiple GS7K systems can then be aggregated to hit 100s of GB/s of performance and PBs of capacity to minimize overall TCO.
► Provides rich feature set of enterprise-class capabilities; migration between SSD, SAS, SATA tiers, Cloud and tape under one global namespace; and federation of GS7K namespaces for multi-site ingest, distribution and collaboration across multiple GS7K Clusters
5
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Challenges
► Traditional NAS scale-out storage systems aren't scaling in performance to meet big data analysis needs• Scale-out NAS only goes to few GB/sec• Customers in Big Data markets need more
► Parallel file systems are hugely scalable, but...• Give up many critical enterprise data management capabilities; or,• Are complex to install/manage/optimize and difficult to use
► Customers with large, business valued data sets need:• Performance of a Parallel File System• Feature Rich• Simplicity of a scale-out appliance• Option of NAS Connectivity
6
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Introducing the GS7K7
Easy-to-deploy, High Performance Scale-Out Parallel File System Appliance
Performance Optimized
Feature Rich
Scalable
• Power of the SFA Architecture combined with the performance and scalability of IBM’s GPFS file system
• 12 GB/sec per scale-out building block
• Enterprise Class Features for Big Data
• Cloud Connectivity• Data Tiering• Integrated Backup using
Tivoli
• Start Small – 4U/60 Drives
• Scale out to 100s of GB/sec of performance and 100s of PBs Capacity
• Building block approach
Simple
• All-in-one appliance with processing and storage pre-loaded
• Power of a Parallel File System in a simple to use appliance
• Singe Vendor Support
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
2PB2PB4PB4PB6PB6PB8PB
Scalable Building Block for Big Data(Animated – please see in full screen mode)
Big Data
Add additional Appliances to Linearly Scale
Performance and Capacity
12GB/s12GB/s24GB/s24GB/s36GB/s36GB/s48GB/s48GB/s
GS7K
Single Global Namespace
…….
Integrate multiple appliances to scale to over 100s of GB/s & 10’s of PBs
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
DirectMon™ | Simple, Scalable Appliance UI
Storage Management Made Simple► A powerful, intuitive single
pane of glass to monitor &manage the GS7K Appliances
► Web interface delivers 100% of monitoring capabilities anywhere, any time
► Performance and Capacity Metric Monitoring: Monitor over 350 metrics to model, problem solve and predict
► Provision and configuration tool sets► User Defined Alerts and Thresholds
Get Up and Running in a matter of Hours
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
10 GS7K - Target Markets
Life Science
Financial
Oil and Gas
Manufacturing
Web 2.0
Social Networks
Cloud Computing
Intelligence
Government
Surveillance
Broadcast
Video on Demand
Special Effects
But NOT Designed for Data Center Applications like Oracle, Exchange, SAP, CRM, ERP
applications Not for Virtualization, VDI etc
Government & Academia
Professional
MediaCloud, Web, &
Telco InfrastructureHPC & Enterprise Big Data Analysis
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
GS7K Features - At-a-glance
Data Protection/Mgmt Intelligence
11
Snapshots and Rollback
Integrated Backup
Mirroring & Aync (TSM) ReplicationI*
Cloud Connectivity and Tiering*
Non Disruptive Scaling, Restriping, Rebalancing
Multi-Tiered File System w/ HSM*
DirectMon Single Pane of Glass
Data Delivery Intelligence
ReACT Cache Management
SFX Storage Cache
Read Quality-Of-Service Engine
Storage Infrastructure Intelligence
World’s Densest – 84 drives in 4U
DirectProtect™ Self-Healing Drive Protection
Data Access Options
Parallel File System (WINDOWS & LINUX), NFS, CIFS
InfiniBand – FDR 56 Gb/s
*Optional, Additional Licenses Required
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Manage Data Intelligently
Information Lifecycle Management*► Built in ILM
Policy based Data Migration
Seamless integration with Tivoli Storage Manager (TSM) to migrate data to and from Tape
► All data is instantly accessible and managed from a single point.
Automate migration between SSD, SAS, SATA and Tape
Single Global Namespace
Automatically
HSM To Tape
Cloud Tiering
Active Tier
SAS Tier
SSD TierPolicy driven
SATATier
Policy driven
Cloud Tiering*► Tier, Archive and Migrate to a WOS cloud
o ILM-based Rules, executed periodically or can be scheduled
► Offload GS7K files to WOS to free up space & improve performance
► View archived files from GS7K as a WOS (NFS) mount point
*Optional, Additional Licenses Required
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Collaborate, Distribute and Federate
Site 1
Site 2
Compute
Grid
Overview► Federation of GS7K namespaces,
using WOS as the wide area distribution mechanism
► Multi-site ingest, distribution & collaboration capacity for up to 8 GS7K Clusters
Access Mechanisms► Data is ingested and written into
GS7K & copied to WOS, or written directly to WOS – and then exposed as a NAS mount point
► Data is ingested into WOS via NFS or CIFS and accessible to GS7K as a NAS mount point
Compute
Grid
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
14 Backup Slides
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
GS7K | Building Blocks15
Base Enclosure
Base + 1
Base + 2
Base + 3
Base + 4
Number of Drives
60 144 228 312 396
Performance 6 GB/s 11 GB/s 12 GB/s 12 GB/s 12 GB/s
Rack Space 4U 8U 12U 16U 20U
Scale Out to Any Performance and Capacity Requirements using these Building Blocks
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
Peace of Mind
Snapshots
Replicated Data and Metadata
Flexible RAID
Tape Backup*
DirectProtect
Cloud Connect & Archive
Data Protection at Multiple Levels► Policy-driven snapshots protect against accidental
deletion, corruption or viruses
► Synchronously replicated data and metadata adds reliability
► Cloud Tiering and Archival
► Flexible RAID configurations provide parity protection against disk failures
► Integrated backup (with Tivoli Storage Manager*) uses an optimized journal to efficiently backup changed data
► DirectProtect to automatically detect and correct silent data corruption
*Optional, Additional Licenses Required
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
GS7K SpecificationsSpecificationsProtocols NFS v3, CIFS SMB 1.0, 2.0 and 2.1
Linux and Windows Parallel FS Clients
Performance 12GB/sec per scale-out building block
Metadata Distributed
Storage Host Ports: 4 x 56Gb/s FDR InfiniBand Ports
Drive Support SSD, Performance SAS and Capacity SAS/SATA drives, 3.5” 60 in base unit, 396 with expansion enclosures
File Servers per base enclosure
4 NSDs
Data Protection • RAID 1/5/6• Up to 256 Snapshots per Volume – policy driven• Snapshot Rollback• DirectProtect & DirectProtect+• Online & Automatic Storage Rebalancing• High Speed Defragmentation
Replication • Synchronous Data and Metadata• Asynchronous using WOS
Data and System Management
• Quotas (users, groups, file sets)• DirectMon Centralized configuration and
monitoring solution• Optional HSM Interface for ILM (IBM HPSS)• Optimized Backup (IBM TSM)
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
DirectMon™ | Monitoring
Selection Tree
Shows NSDs and Storage
Select a Component to Drill Down
Aggregated Performance and Capacity Metrics
System Health for All NSDs and Storage in the GS7K Cluster
High-Level GS7K Monitoring
NSD Specific Monitoring
Selection of Monitoring Views Available for GS7K NSDs
Capacity Monitoring
Health Status
Time-Based View Options
Graphic Monitoring Report
ddn.com© 2014 DataDirect Networks, Inc. * Other names and brands may be claimed as the property of others.• Any statements or representations around future events are subject to change.
DirectMon™ | Simple, Scalable Appliance UI
Storage Management Made Simple► A powerful, intuitive single
pane of glass to monitor &manage the GS7K Appliances
► Web interface delivers 100% of monitoring capabilities anywhere, any time
► Performance and Capacity Metric Monitoring: Monitor over 350 metrics to model, problem solve and predict
► Provision and configuration tool sets► User Defined Alerts and Thresholds