Published on December 12, 2013
Introducing StorNext 5 Appliances
Modern Workflows Have Changed Operations Video, Image and Sensor data paths are 100% digital Deliver Globally distributed teams require shared access Various sized unstructured data files are managed Topologies and interconnects have evolved
StorNext 5 Appliances A New Era for Digital Workflows Built from the Ground Up for – Performance – Scalability – Flexibility
Engineered from the Ground up Next Generation Technology A New Modern Core – New cache management – Multiprocessor aware with lower CPU utilization – More compact on-disk format – New reporting/management framework
5x Greater Scalability Up to 5 Billion Files – Metadata scalable to petabytes – Designed for large block allocation space – No slowdown as FS grows and ages
10x Faster Performance* Across the Board Performance Improvements – All files sizes improve – File creates and deletes up to 10x faster – Small file performance greatly increased up to 100% – Takes full advantage of SSD Flash and multi-core processors * Performance measured in standardized testing against StorNext v4.2 file create and delete performance, real-world application performance may vary
Beyond the SAN IP/NAS + SAN and beyond Topology Agnostic – Fibre Channel, IP/NAS and beyond - all highly optimized with new cache parameters and better system tuning – All operating systems enjoy better performance and lower latency – Takes advantage of InfiniBand for lower latency applications such as HPC
Optimized for StorNext Q-Series and Lattus Object Storage Improved Affinities Provide Better Disk Performance – Automated tuning and reduced parameters, now self-tuned for Quantum disk – Buffer cache I/O improvements and smaller block allocation improve performance on Petabyte-scale Lattus Object Storage
Native LTFS Capability Choice of LTO Tape Format LTFS format – – – – – Meets LTFS standards Non-proprietary format Content delivery Collaboration Long-term vaulting Quantum ANTF format – Higher read and write performance – Single file can span multiple tapes
Architected for the Long-Term Redesigned To Boost Efficiency and Performance – RAM, CPU cache, controller and disk IO all important in FS performance – Shrinking metadata makes all these more effective – Fit more in RAM and cache – Move more with bandwidth – Maintain consistent performance over time – Seamless, non-disruptive software updates
Flash & InfiniBand Optimized Takes Full Advantage of SSD and Flash for Performance – Low latency and high I/O applications benefit most – Improves HPC and analytic operations – Improves small file performance – Support for InfiniBand
Engineered from the Ground up 5x Greater Scalability 10x Faster Performance* Beyond the SAN IP/NAS + SAN and beyond Optimized for StorNext Q-Series and Lattus Object Storage Native LTFS Capability Architected for the Long-Term Flash & InfiniBand Optimized
StorNext 5 Use Cases by Industry End-to-End Content Management Media & Entertainment Real-Time Mission Support Sequence Genes to Cure Disease Intelligence & Surveillance Life Sciences Analyze Data to Accelerate Discovery Natural Resources
Integrated with Workflow Applications 15
Leading Companies Choose StorNext
StorNext 5 Appliance Family End-to-End Workflow Ingest Work in Process Delivery Nearline Long Term Off-site StorNext LTFS StorNext Q-Series High-Performance Disk Storage Lattus Object Storage StorNext AEL Archives
StorNext 5 Availability StorNext 5 will initially be available on StorNext appliances in 2013 StorNext 5 software-only solutions will deliver in the first half of 2014
NEXT-GENERATION OBJECT STORAGE
Data Challenges Converge on Your Business 19
Data Value Drives Changes to Workflows Traditional Infrastructure New Infrastructure Never-Ending Disk Pools Fast Cache Cache for immediate work Smart Data Movement Support unpredictable, on-demand access Legacy Backup & Archive Vault 20 Policy-based, native, unlimited scalability, low TCO
Petabyte Scale Challenges Traditional Architecture Efficiency Traditional RAID is inefficient at Petabyte Scale 88% 75% Percentage usable 66% 100TB 1PB 10PB 10 Disk RAID* Time (days) RAID suffers from unacceptable rebuild times as drive size increases 3 2 1 1TB 21 3TB 4TB *2013 Evaluator Group, Inc.
Traditional Storage Approaches Reduce Data Usage Efficiency Traditional block addressing cannot match scale, access or durability needs Traditional file hierarchies inhibit required global access flexibility 22
Next-Generation Object Storage Matches Modern Data Usage Objectives Unlimited Scalability Independent of physical location Flexible Data Usage No hierarchy – 100% flat namespace Long-Term Online Durability Forward Error Correction (FEC) for data integrity Efficiency allows multiple copies of encoded data distributed across storage nodes Global Data Access Distributed data leveraging global networks 23
Object Storage Provides Immediate Value Cost-Effective Global Data Availability Rapid Access to Historical Data Infinite Scalability & Data Durability 24 Transparent Cloud Migration Big Data Analysis Enabled Content Re-Monetization Made Easy
Lattus-M Object Storage Disk Archive Forever disk archive StorNext-managed policy-based tiering Leverages next-generation object storage Lower latency than tape More efficient than mirrored RAID Includes Storage Manager license 1 Billion Files (per MDC) Archive starts at 200 TB Virtually unlimited scalability
THANK YOU © 2013 Quantum Corporation. Company Confidential. Forward-looking information is based upon multiple assumptions and uncertainties, does not necessarily represent the company’s outlook and is for planning purposes only.
In this slidecast, Skip Levens from Quantum describes the company StorNext5 family of Appliances and Quantum Lattus Object Storage for Big Data ...
In this slidecast, Skip Levens from Quantum describes the company StorNext5 family of Appliances and Quantum Lattus Object Storage for Big Data management.
Introducing Janus and Cardea; of 12
Lattus - Object Storage ... Introducing StorNext5 Appliances ... Coffee with Quantum Prospecting for Lattus M Uncovering Object Storage ...
Introducing StorNext5 and Lattus. Introducing BlogLevel and TweetLevel. Volunteer Co-ordinators’ Network: Issues, Solutions and Support Needs. Login or Join.
SAM introduce Go!, una suite de producción remota de edición de líneas de tiempo. Usuarios de computadoras de escritorio, portátiles y tabletas
In this podcast, Jason Fedder, Ritu Kama, and Ted Willke from Intel describe the company’s new Intel Data Platform, a family of software for Big Data ...
In this video from the ISC Big Data'14 Conference, DK Panda from Ohio State University presents: Performance Optimization of Hadoop Using InfiniBand RDMA.