Leading distributed fusion architecture

The H3C ONEStor storage system adopts a fully distributed architecture: distributed management cluster, distributed hash data distribution algorithm, distributed stateless client, distributed cache, etc; This architecture provides strong guarantees for the reliability, availability, automatic operation and maintenance, and high performance of storage systems. ONEStor distributed storage system can provide various storage interfaces such as blocks, objects, files, etc. on a single platform, providing users with a variety of storage services that can be used as needed.

Linear scalability

The ONEStor system adopts a distributed system design, and its storage capacity and performance increase linearly with the increase of server nodes and hard drives. The system can easily expand to thousands of nodes and EB level capacity, meeting the needs of cloud business expansion.

deduplication

ONEStor distributed storage system provides duplicate data deletion function, which removes duplicate data blocks through fingerprint calculation and comparative analysis of data. Only one copy of duplicate data is retained on the storage medium, and other duplicate copies are indexed. The duplicate data deletion function effectively reduces data redundancy and saves storage space.

Flexible partitioned storage

The ONEStor distributed storage system can achieve mixed deployment of different hard disk types and partition storage. The types of hard drives supported by the ONEStor system include: full HDD, SSD+HDD hybrid networking, and full SSD. In the SSD+HDD hybrid networking mode, the ONEStor system can use SSDs as cache or place SSDs and HDDs in different pools for tiered storage. This hybrid deployment approach can leverage the advantages of SSD's IOPS and throughput, as well as the capacity and price advantages of HDD, making it currently the most reasonable and widely used method.

Rich enterprise level functional features

Multi replica technology ensures data reliability. Under a typical 3-copy configuration, user data will only be truly lost if all 3 copies of data are lost simultaneously. Support online modification of the number of copies.

Support k+m erasure codes, multiple redundancy schemes, effectively reducing TCO.

In the case of a 4+1 erasure code method, if any server fails or crashes, the data will not be lost, and the system can recover the lost data through the original data and verification data.

Fast data reconstruction: ONEStor supports parallel and fast fault handling and reconstruction, with data blocks and replicas scattered throughout the entire resource pool. After a hard disk failure, it can automatically reconstruct within the resource pool, with a data reconstruction time of up to 30 minutes/TB.

Dual active storage: Supports remote dual active function through distributed clusters to achieve dual active storage of data from two sites. The host can read and write to the same dual active volume concurrently, and any equipment failure on either side of the site will not affect the operation of the upper layer business system.

Strong data consistency: The replicas have strong consistency, and confirmation returns are only made after all replicas have completed writing, placing the reliability of user data first in the entire storage system and ensuring the highest reliability of user data.

Flash acceleration capability: Through a unique hot and cold data layering mechanism, hot data is stored in a multi-level cache layer composed of memory and SSD, greatly improving the probability of data hits and enhancing read efficiency.

Comprehensive data protection and disaster recovery capabilities, such as snapshots, volume copies, QoS policies, remote replication, fault domains, protection domains, etc.

Automated operation and maintenance capabilities

Automated and rapid deployment of storage clusters, including batch deployment, single node addition and removal, single disk addition and removal, etc., accelerates the system's online speed.

The automatic monitoring and alarm system can quickly identify and troubleshoot problems in case of malfunctions, improving operational efficiency.

End to end monitoring capability, providing comprehensive performance and health status monitoring from physical servers to storage pools, including monitoring of CPU, memory, network cards, read/write IOPS, read/write bandwidth, response time, and component health status.

According to different hardware capabilities, nodes in the cluster can be flexibly configured and deployed to improve hardware resource utilization.

Convenient fault domain isolation and flexible selection of data storage locations to improve business application reliability. Support the design of three-level fault domains for hosts, cabinets, and data centers, and automatically generate cluster device topology after specifying node roles during the deployment phase.

Automated data balancing capability ensures high availability of the cluster by automatically recovering data in the event of node additions, deletions, or failures.

The ability to predict disk failures and storage usage, based on advanced AI and big data algorithms, predicts the lifespan and storage usage of hard disk devices. When the system predicts a possible disk failure, it automatically generates warning information to notify administrators to replace storage media in a timely manner; Intelligent prediction and alarm of the remaining disk capacity that can support usage time, with customizable alarm thresholds.

Complete log management capabilities, administrators can customize the time span to export logs, including cluster software and hardware adjustment information, network operation status, heartbeat status, storage pool status changes, user behavior records, etc.

Comprehensive auditing capabilities, supporting the control of three separate roles, ensuring system access security. The audit content covers all system operation records, and the storage console supports SSL access encryption.

A friendly management interface that intuitively displays the topology status of the cluster, including device information such as data centers, racks, servers, network cards, component abnormal events, monitoring history, etc.

Good compatibility

Supports servers from mainstream X86 manufacturers such as Dell/Inspur/H3C/HPE/Huawei, and can be adapted to X86 servers according to requirements.

The storage resource pool supports heterogeneous servers and mixed disk insertion.