Next-Gen Storage for High-Performance Data Analytics

High-Performance Computing (HPC) is already being extensively used in areas such as oil and gas exploration, weather forecasting, and Computer-Aided Engineering (CAE) simulations. Today, as it converges with big data, traditional HPC is evolving towards High-Performance Data Analytics (HPDA). As such, heterogeneous computing power and data is experiencing explosive growth, and the need for multi-protocol data access is becoming ever more crucial.

Huawei is dedicated to meeting these challenges head-on, by building the next generation of storage for HPDA, to help organizations like yours fully unleash the potential of data, with hybrid workloads-oriented capabilities, a high-density design, and multi-protocol interworking.

Hybrid Workloads-Oriented

Each rack unit in a storage system supports both a high bandwidth of 32 GB/s, as well as 400,000 Input/Output Operations Per Second (IOPS), delivering industry-leading performance density.
High-Density Design

The design features high-density capacity hardware with 24 disks/U, high-density Non-Volatile Memory express (NVMe) all-flash hardware with 16 Solid State Drives (SSDs)/U, and intelligent tiering of hot and cold data.
Multi-Protocol Interworking

Seamless multi-protocol interworking between the Portable Operating System Interface (POSIX), Network File System (NFS), Common Internet File System (CIFS), Hadoop Distributed File System (HDFS), S3, and MPI-IO, without data copying, in order to provide smooth HPDA evolution.

HPC Storage Solution Application Scenarios

Oil & Gas Exploration
One System for Multiple Phases in Oil & Gas Exploration

◆One-time construction and any-time capacity expansion: Optimize investment with a fully symmetric architecture, small-scale initial deployment, and linear expansion to thousands of nodes in response to data growth.

◆One storage device with automatic tiering based on data access frequency: Automatic tiering of hot, and cold data, with no impact on services and reduced Total Cost of Ownership (TCO) throughout the lifecycle.

◆N sets to one set, migration-free, and easy maintenance: The same architecture supports seismic data processing, interpretation, and reservoir simulation, eliminating the need for data migration, improving exploration efficiency, and reducing data copies.
Genome Sequencing
Reduced TCO with Automatic Tiering of Hot, and Cold Data

◆Automatic tiering of hot, and cold data: Free mobility of hot, and cold data between storage media without affecting ongoing services greatly lowers the TCO.

◆Migration-free, with one data copy throughout the process: The same data copy is accessed over different protocols in sequencing, alignment, analysis, and display phases.

◆Analytics performance improves with all-ﬂash + private clients: NVMe all-ﬂash storage with 160 GB/s bandwidth per chassis and more than 3 GB/s single-thread bandwidth for the Distributed Parallel Client (DPC).
Weather Forecasting
Parallel Computing for Higher Prediction Accuracy and Speed

◆Multiple phases in one storage system: Collection, preprocessing, mode calculation, post-processing, and data display are served by one storage system, eliminating the need for data migration.

◆Data tiering and archiving without affecting ongoing services: Automatic tiering of hot, and cold data, with no impact on services; reduced TCO throughout the lifecycle.

◆Parallel analysis of mass data: Support for Message Passing Interface-Input/Output (MPI-IO), enabling multiple compute nodes to access the same ﬁle and greatly improving the efficiency of analysis.
Supercomputing
Multi-Protocol Interworking and Hybrid Workloads for Multiple Tenants

◆Adaptation to complex workload models: One storage system supports high bandwidth and IOPS, while DPCs support MPI-IO, providing high single-thread and single-client performance.

◆Flexible data access by multiple applications: Interworking between ﬁle, HDFS, and object protocols makes HPDA data access more efficient, and eliminates the need for data migration.

◆Data and load isolation between tenants: Multi-tenant and Quality of Service (QoS) capabilities isolate service resources and performance.

HPC Storage Solution Success Stories

Exploration Company Improves Storage Architecture for Success

Data is one of the most valuable resources in a smart world, and the volume of data collected in the oil and gas industry is growing exponentially.

Learn More

Learn More About Our Products

OceanStor Pacific Series Storage

Designed for massive data scenarios, these distributed storage systems provide diversified storage services for virtualization and cloud resource pools, big data analytics, HPC, and backup archiving, helping enterprises unleash the value of massive data.

Learn More

Partners

If you are already a Partner, please click here to get more marketing information. Please click here to enter the Partnership area for project reporting, order management, technical support, and other operations.

Login

Huawei OceanStor Pacific Series

OceanStor Pacific Storage for HPDA