Anfragen
OceanStor 100D
Huawei OceanStor 100D — originally branded as FusionStorage — is an intelligent, distributed storage product developed in-house, offering scale-out capabilities to support the business needs of tomorrow, today. It provides industry standard interfaces, including file, Hadoop Distributed File System (HDFS), and object storage, for upper-layer applications. This greatly simplifies operations complicated by siloed storage systems, offering ceaseless, uninterrupted operations for services, highly efficient processing for diverse data, and more economical storage for mass data.
File storage is applicable to High-Performance Computing (HPC) and other performance-intensive scenarios with full compatibility for native Network File System (NFS) and System Bay (SMB) protocols, as well as Portable Operating System Interface (POSIX) and Message Passing Interface (MPI).
HDFS storage provides a decoupled storage-compute big data solution using native HDFS for on-demand resource configuration, slashing Total Cost of Ownership (TCO) and delivering a superior user experience. In addition, HDFS storage can coexist with legacy coupled storage-compute architecture and has been widely adopted for big data applications in finance, Internet log retention, e-government, and Smart City projects.
Object storage enables a single bucket to house up to 100 billion objects without compromising performance, thereby eliminating multiple buckets and application changes for large-scale applications. It automatically tiers data to the proper media, avoiding data migration and reducing space used. Object storage is extensively used in scenarios involving production, storage, backup, and archiving of bank check images, audio and video recordings, medical images, Business to Business (B2B) electronic documents, and the Internet of Vehicles (IoV).
OceanStor 100D uses End to End (E2E) reliability assurance at the Input/Output (I/O), device, system, and data center levels to offer highly available storage services for complex enterprise applications. Specifically, OceanStor 100D supports E2E data integrity check and self-healing. It provides comprehensive sub-health detection and self-healing for disks, nodes, and networks. System reliability remains uncompromised even if four nodes or four cabinets fail simultaneously.
By using key technologies such as FlashLink® performance acceleration and Artificial Intelligence (AI) Fabric lossless networking, OceanStor 100D ensures performance remains stable even when 100 billion objects are hosted in a single bucket. This means that even massive amounts of small files can be accessed with ease. OceanStor 100D also implements efficient data usage and converged management of multiple storage services, such as file and object. Furthermore, both old and new generations of hardware, and multiple platform hardware products are supported, enabling intelligent system management throughout the entire lifecycle.
Each OceanStor 100D cluster supports a large quantity of nodes, from 3 to 4096, enabling on-demand and elastic scalability. Performance grows linearly as capacity increases, preserving current investments made. Meanwhile, elastic Erasure Coding (EC) for data redundancy protection delivers 91.6% of disk space utilization*, 2.75x that of the traditional three-copy mode. Moreover, abundant primary storage tiering policies achieve optimal storage and management of hot, warm, and cold data.
| Model |
OceanStor 100D |
||
| System Architecture |
Fully distributed architecture |
||
| Storage Service Type | File Storage | HDFS Storage | Object Storage |
| Supported Protocols | NFS, CIFS, POSIX, MPI | HDFS | Amazon S3, NFS |
| Data Redundancy Protection | EC | EC | EC |
| Key Features |
• Load balancing • Quotas • Data tiering • Snapshots • QoS • DIF • Distributed Parallel Clients (DPCs) • AD, LDAP, NIS domain authentication |
• Append, rename, flush, hFlush • Multiple namespaces • Kerberos and LDAP authentication • Directory and namespace quotas • Small-file aggregation • Data tiering • DIF • Recycle bins • Coexistence with the existing local HDFS |
• Unified namespace, multiple tenants and resource pools • Small-object aggregation • QoS • Data lifecycle management • Data consistency checks • Synchronous replication • Asynchronous replication • Object deduplication • WORM • LDAP/AD authentication |
| Front-End Network Types | • 10 GE, 25 GE, 40 GE, or 100 GE TCP/IP • 10 GE, 25 GE, 40 GE, or 100 GE RoCE • 100 Gb/s InfiniBand |
10 GE or 25 GE TCP/IP | 10 GE or 25 GE TCP/IP |
| Internal Network Types | • 25 GE or 100 GE RoCE • 100 Gb/s InfiniBand |
10 GE or 25 GE RoCE | 10 GE or 25 GE TCP/IP |
| Compatible Platforms and Components | • OpenMPI, MPICH2, Huawei MPI • Kubernetes |
Mainstream big data components: MapReduce, Hive, HBase, Spark | Backup software: Commvault, NBU, Atempo, Veeam, Rubrik |
| Recommended Node Models |
• Performance model (2U): P100 • Capacity mode (4U): C100 |
• Performance model (2U): P110, P100
• Capacity mode (4U): C110, C100 |
• Performance model (2U): P110, P100
• Capacity mode (4U): C110, C100 |
*：For more information, please contact your local Huawei sales representatives.
China Merchants Bank: OceanStor Distributed Storage Hosts Critical and Emerging Workloads
China Mobile: OceanStor Distributed Storage Accelerates Core Workload Upgrades
