OceanStor Scale-Out Storage

Efficient Processing for Diverse Data

Software algorithms are offloaded onto processors for hardware acceleration. In addition, OceanStor scale-out storage uses an intelligent and lossless network that delivers consistent latency of less than 1 ms¹, quickly responding to critical workloads.

A single bucket houses up to 100 billion objects without compromising performance, eliminating the need to use multiple buckets for large applications.

Excellent Input/Output (I/O) throughput is provided for applications that involve a massive volume of small files. Fully symmetric distributed architecture enables massive and linear performance expansion.

Full-lifecycle storage system management predicts and handles resource usage and faults in advance, radically streamlining Operations and Maintenance (O&M).

Notes: ¹ Applicable to scale-out block storage.

Uninterrupted Operations for Online Services

State-of-the-art distributed active-active architecture across clusters² implements load balancing, where RPO = 0 and RTO ≈ 0, protecting core data assets.

The innovative Disaster Recovery (DR) solution extends across 2–12 sites³, offering a choice between cross-site Erasure Coding (EC) and multi-copy options, delivering improved reliability and space utilization.

EC provides powerful redundancy protection to guarantee the stable running of applications even if four nodes or four cabinets are simultaneously faulty.

Comprehensive system sub-health detection and pre-processing accurately eliminate potential network, node, and component faults — all in advance.

End to End (E2E) data consistency verification provides high data availability for concurrent services.

Notes: ² Applicable to scale-out block storage. ³ Applicable to scale-out object storage.

Economical Storage for Mass Data

The OceanStor scale-out storage series implements EC to deliver industry-leading disk space utilization of 91.6%. That's 2.75x the figure of a traditional three-copy mode, slashing the hardware needed and the physical space required, as well as bringing down energy consumption.

The series provides state-of-the-art data reduction solutions for life sciences and digital pathology, optimizing the use of storage space.

The series also scales out to 4096 nodes, for flexible resource purchasing and deployment on demand.

It provides precision purchasing with a sales model of available capacity⁴, optimizing investments made.

Equipped with native Hadoop Distributed File System (HDFS) capabilities, the decoupled storage-compute big data solution is compatible with mainstream big data platforms and components, for diversified data analytics with 30% lower Total Cost of Ownership (TCO).

Notes: ⁴ Actual capacity that stores service data.