PLStor D Series Distributed Storage

Diversity data is used more efficiently

The PLStor D series uses a series of performance acceleration technology innovations such as directory DHT partitioning, intelligent hard disk granularity management, large U/O direct pass-through with small U/O aggregation, and multi-level intelligent cache caching to achieve a set of storage that supports high-bandwidth and high-OPS business access requirements. Among them, the new generation of parallel file systems supports both NFS and CIFS standard protocol access, and can provide POSIX and MPI-IO access capabilities through DPC (Distributed Parallel Client) distributed parallel clients. DPC can connect to multiple storage nodes at the same time to achieve I/O-level load balancing, and support RDMA access to achieve higher single-stream and single-client performance.

The PLStor D series optimizes the abundant computing power provided by high-performance processors, unloads some storage functions to the processor layer, and realizes the hardware acceleration of software functions. Compared with other products with the same configuration, it can achieve a 20% increase in storage computing power. In the SPEC SFS2008 test, the throughput of 10 nodes can reach 500,000. Whether your data center needs to expand IO-intensive, bandwidth-intensive, latency-sensitive or large-capacity demand services in the future, the PLStor D series can carry them on demand.

More stable load carrying for complex business

PLStor D series supports end-to-end data integrity verification and self-healing. It can solve silent data errors such as byte jump, read bias, and write bias through online verification and background periodic verification to ensure high data availability. The powerful elastic erasure coding (EC) data redundancy protection technology allows a single system to tolerate up to 4 nodes at the same time. It supports dynamic EC. When a node fails, it automatically adjusts the EC ratio to ensure that the system reliability is not degraded. At the same time, through a series of technologies such as multi-module concurrent business takeover, it ensures that the service switching time of a single node failure can be as low as 10 seconds. It provides comprehensive sub-health detection and self-healing for hard disks, nodes, networks and other dimensions, and monitors the device status in real time. It builds a disaster recovery system through replication or distributed active-active technology to achieve data center-level high availability. The IWO-level, system-level, and data center-level end-to-end reliability assurance technology provided by PLStor D series provides solution-level high availability of up to 99.9999%, helping you build data protection solutions that meet different business protection needs.

Massive data storage costs are better

The PLStor D series organizes storage media such as HDD and SSD into a large-scale storage resource pool through distributed technology, provides industry standard interfaces for upper-layer applications and clients, and eliminates the uneven utilization of hardware resources caused by the chimney-style storage system construction of traditional data centers. The product supports initial small-scale deployment and can be horizontally expanded to a large-scale storage cluster of up to 4096 nodes, providing linear growth capabilities in performance and capacity, and optimizing current investments.

PLStor D series can use the leading elastic EC data redundancy protection mechanism and copy protection mechanism. Compared with the traditional multi-copy method, EC technology increases the hard disk space utilization rate by nearly 2 times; at the same time, it provides a variety of redundancy ratios for flexible selection and on-demand deployment, and can support a large EC ratio of up to 22+2 and a hard disk space utilization rate of up to 91.6%, helping you save hardware investment. Based on a rich main memory classification strategy, it supports automatic data migration between high-performance pools and large-capacity pools, realizing more reasonable storage and management of cold and hot data.

At the same time, the PLStor D series also provides scenario-based solutions to help users further optimize IT investments. For example, the big data computing and storage separation solution can improve data analysis efficiency, and the intelligent management function supports the management of third-party HDFS, which can protect existing investments and reduce TCO to the greatest extent.

Simplified data lifecycle management

The PLStor D series provides unified multi-product and storage service integration management and intelligent operation and maintenance capabilities. Intelligent resource prediction can shield storage resource service risks in advance, helping you to make more accurate decisions on capacity expansion, procurement, and service changes; intelligent fault location provides a comprehensive system sub-health detection and processing mechanism to achieve accurate and rapid fault location and recovery.

Object Storage

Support multiple access protocols such as Amazon S3, NFS, CIFS, POSIX, MPI-IO, HDFS, etc. to access the same object;

Supports the recycle bin function, and can configure the retention time and scanning cycle of objects in the recycle bin;

Supports the recycle bin function, and can configure the retention time and scanning cycle of objects in the recycle bin.

Block Storage

Supports rapid data reconstruction. When a disk or storage node fails, the system can automatically rebuild data. Without manual intervention, the data reconstruction speed can meet the requirement of: ≤15 minutes per TB.

Support intelligent operation and maintenance;

Supports snapshot reconstruction. When the original volume data changes, the snapshot volume data can be refreshed through reconstruction.

Supports SSD wear life identification, early warning and isolation processing.

Supports storage node sub-health function: If the storage node service latency is too high or the memory usage is too high, an alarm will be issued and a solution will be provided.