General computing, heterogeneous computing, collaborative computing, edge computing and other diversified product layout
The PLStor D series organizes the local storage resources of each hardware node through the storage system software, and provides file storage services, big data storage services, object storage services or block storage services for upper-layer applications on demand to achieve higher efficiency in the use of diverse data , The complex business load is more stable and the mass data storage cost is better .
File storage : Compatible with native NFS, CIFS protocol and POSIX , M PI-IO parallel interface, widely used in HPC ( High Performance Computing) and other high-performance storage scenarios .
a big data storage-computing separation solution based on native H DFS semantics and does not require the installation of plug-ins on computing nodes, combined with intelligent grading, coexistence of new and old , etc., to help you reduce the total cost of ownership and provide a consistent user experience.
Object storage : Compatible with Amazon S3, providing excellent small object processing performance and perfect disaster recovery capabilities .
Block storage: Provides SCSI or iSCSI access, with fully distributed Active-Active active - active, cabinet-level redundancy and other high-availability solutions, which are widely used in scenarios such as virtualization /cloud resource pools and databases.
PLStor D series , through a series of performance accelerationtechnology innovations such as directory DHT partitioning, intelligent harddisk granular management, large I / O pass-through combined with small I/Oaggregation, and multi-level intelligent cache caching, can realize a set ofstorage that supports both high bandwidth and high OPS type. business accessrequirements. Among them, the new generation of parallel file system not onlysupports NFS and CIFS standard protocol access, but also provides POSIX andMPI-IO access capabilities through DPC (Distributed Parallel Client )distributed parallel client. DPC can connect to multiple storage nodes at thesame time to achieve I / O-level load balancing, and supports RDMA access atthe same time , which can achieve higher single-stream and single-clientperformance. The DPC distributed parallel client is deployed ,and the product cansupport large-scale computing clusters of up to 5,000 nodes.
PLStor D series optimizes the abundant computing power provided by high-performance processors,offloads some storage functions to the processor layer, and realizes hardacceleration of software functions. Compared with other products with the sameconfiguration , it can achieve a 20% increase in storage computing power . Whetheryour data center needs to expand I/O-intensive, bandwidth-sensitive, latency -sensitiveor large-capacity services in the future, the PLStor D series can carry it ondemand.
PLStor D series supports end-to-end data integrityverification and self-healing, and can solve silent data errors such as bytejumps, read offsets, and write offsets through online verification andbackground periodic verification , ensuring high data availability; powerfulThe Elastic Erasure Coding (EC) data redundancy protection technology enables asingle system to tolerate a maximum of 4 simultaneous node failures; supportsdynamic EC, when a node fails, the EC ratio is automatically adjusted, systemreliability is not degraded, and at the same time , through a series oftechnologies such as multi-module concurrent service takeover, to ensure thatthe service switching time of a single node failure can be as low as 10seconds; comprehensive sub-health detection and self-healing for hard disk,node, network and other dimensions , real-time monitoring of equipment status;through replication or Distributed active- active technology builds a disasterrecovery system to achieve data center-level high availability. The I/O-level,system-level, and data center-level end-to-end reliability assurancetechnologies provided by the PLStor D series provide solution-level highavailability up to 99.9999%, helping you build data protection solutions thatmeet different business protection needs.
The PLStor D series organizes storage media such asHDD and SSD into a large-scale storage resource pool through distributedtechnology, provides industry-standard interfaces for upper-layer applicationsand clients, and eliminates the uneven utilization of hardware resources causedby the construction of traditional data center chimney storage systems. problem. The product supports initial small-scale deployment, and can scalehorizontally to a large - scale storage cluster of 4,096 nodes, providinglinear growth of performance capacity and optimizing current investment .
PLStor D series adopts the leading elastic EC data redundancy protection mechanism .Compared with the traditional multi -copy method, EC technology improves theutilization of hard disk space by nearly 2 times ; It can support a large ratioof 2 2 + 2 E -C at most, and the utilization rate of hard disk space is as highas 9 1.6 % , helping you save hardware investment . Based on rich main memorygrading strategies, it supports automatic data migration betweenhigh-performance pools and large-capacity pools, enabling more reasonablestorage and management of hot and cold data.
At the same time, the PLStor D series also provides scenario -based solutions tohelp users further optimize IT investment. For example, the big data computingand storage separation solution that can improve the efficiency of dataanalysis , provides intelligent management functions to support the managementof third-party HDFS , maximizes protection of existing investment and reduces TCO.
The PLStor D series provides unified multi-productand storage service integrated managementand intelligent operation and maintenance capabilities . Intelligent resourceprediction can shield storage resource service risks in advance and help youmake more accurate decisions on capacity expansion, procurement, and servicechanges ; intelligent fault location provides a comprehensive system sub-healthdetection and processing mechanism to achieve accurate and rapid fault locationand recovery; at the same time, Compatible with a variety ofhardware ecosystems, it supports co-cluster management of x86 architecture andKunpeng (ARM architecture) processor storage nodes.a
| Product number | PLStor D 28520 | PLStor D 28540 | PLStor D 28550 | PLStor D 28920 | PLStor D 28950 |
|---|---|---|---|---|---|
| system structure | Fully symmetric distributed architecture | ||||
| Maximum raw capacity per chassis | 216TB | 648TB | 2160TB | 92TB | 614.4TB |
| Height of each case | 2U | 4U | 5U | 2U | 5U |
| Nodes per chassis | 1 | 1 | 2 | 1 | 8 |
| Maximum number of main storage disks per node | 12 | 36 | 60 | 12 | 10 |
| processor per node | 2 Kunpeng 920 processors or 2 x86 architecture processors | 2 Kunpeng 920 processors or 2 x86 architecture processors | Kunpeng 920 processor | 2 Kunpeng 920 processors | Kunpeng 920 processor |
| Maximum memory per node | 256GB, 512GB | 256GB, 512GB | 256 GB | 512 GB | 256 GB |
| Maximum cache per node | 4 NVMe SSDs | 4 NVMe SSDs | 4 Half-Palm NVMe SSDs | not involving | not involving |
| system disk per node | 2 x 600 GB SAS HDDs or 2 480GB SATA SSDs | 2 x 600 GB SAS HDDs or 2 480GB SATA SSDs | 2 x 480 GB SSDs | 2 x 600 GB SAS HDDs or 2 480GB SATA SSDs | 2 x 480 GB SSDs |
| data disk type | 3.5-inch HDD | 3.5-inch HDD | 3.5-inch HDD | 3.5" SAS SSD | Half-Palm NVMe SSD |
| Front-end businessnetwork① | l 25GE or 100GE RoCE l 100 Gb/s EDR InfiniBand | l 10 GE, 25 GE or 100GE TCP/IP l 25GE or 100GE RoCE l 100 Gb/s EDR InfiniBand | l10 GE, 25 GE or 100 GE TCP/IP l25GE ,100GE RoCE l100Gb/s EDR/HDR InfiniBand | l 10 GE, 25 GE or 100 GE TCP/IP l 100 GE RoCE l 100 Gb/s EDR InfiniBand | l100 GE RoCE l100 Gb/s EDR/HDR InfiniBand |
| l 10GE , 25GE or 100GE RoCE _ l 100 Gb/s EDR InfiniBand | l 10GE , 25GE or 100GE RoCE _ l 100Gb / s EDR InfiniBand _ | 25 E RoCE | l 25GE _ _ or 100 GE RoCE l 100Gb /s EDR InfiniBand | ||
| Data redundancy protection mechanism | Erasure Coding : Support N+M redundancy protection, M supports 2, 3 or 4 | ||||
| storage access protocol | NFS, CIFS, POSIX, MPI-IO, HDFS, Amazon S3, etc. | ||||
| Key Features | SmartQuota, SmartTier, SmartQoS, SmartEqualizer, SmartMulti - Tenant, SmartEncryption, SmartAuditlog,HyperLock,HyperSnap, HyperReplication (A),SmartIndexing,Recycle Bin, SmartInterworking, DIF, Object Versioning②, SmartTakeover③ | ||||
| data self-healing | Automatic parallel reconstruction, efficiency up to 2TB/hour | ||||
| Chassis size (H×W×D) | lKunpeng model: 86.1 mm×447mm×790mm lx86 model: 86.1mm×447mm×748mm | lKunpeng model: 175mm×447mm×790mm lx86 model: 175mm×447mm×748mm | 219.5mm x 447mm x 1030mm | 86.1 mm ×447mm×790mm | 219.5mm×447mm×926 mm |
| Maximum weight per chassis (including hard drives) | lKunpeng model:≤32 kg lx86model:≤34.1 kg | ≤65kg | ≤164kg | ≤32kg | ≤115kg |
| Working temperature | 5℃～35℃ | 5℃～35℃ | 5℃～35℃ | 5℃～35℃ | 5℃～35℃ |
| Working environment humidity | 8% RH～90% RH, no condensation | 8% RH～90% RH, no condensation | 5% RH～90% RH, no condensation | 8% RH～90% RH, no condensation | 5% RH～90% RH, no condensation |
| Product number | PLStor D 28520 | PLStor D 28540 | PLStor D 28920 |
|---|---|---|---|
| system structure | Fully symmetric distributed architecture | ||
| Maximum raw capacity per chassis | 216TB _ | 648TB _ | 92TB |
| Height of each case | 2U | 4U | 2U |
| Nodes per chassis | 1 | 1 | 1 |
| Maximum number of main storage disks per node | 12 or 25 | 36 | 12, 22 or 25 |
| processor per node | 2 Kunpeng 920 processors or 2 x86 architecture processors | 2 Kunpeng 920 processors or 2 x86 architecture processors | 2 Kunpeng 920 processors or 2 x86 architecture processors |
| Maximum memory per node | 512 GB, 768 GB or 1 TB | 512 GB, 768 GB or 1 TB | 768 GB or 1 TB |
| Maximum cache per node | 4 NVMe SSDs or SAS SSDs | 4 NVMe SSDs or SAS SSDs | not involving |
| system disk per node | 2 x 600 GB SAS HDDs or 2 480GB SATA SSDs | 2 x 600GB SAS HDDs or 2 480GB SATA SSDs | 2 x 600GB SAS HDDs or 2 480GB SATA SSDs |
| data disk type | 2.5" or 3.5" HDD | 3.5-inch HDD | 2.5" or 3.5" SAS SSD, NVMe SSD |
| Front-end business network | l 10 GE or 25 GE TCP/IP l 25 GE RoCE | l 10 GE or 25 GE TCP/IP l 25 GE RoCE | l 10 GE or 25 GE TCP/IP l 25 GE or 100GE RoCE l 100Gb /s InfiniBand |
| storage internet | l10 GE or 25 GE TCP/IP l25 GE RoCE | l10 GE or 25 GE TCP/IP l25 GE RoCE | l10 GE or 25 GE TCP/IP l25 GE or 100 GE RoCE l100Gb /s InfiniBand |
| Data redundancy protection mechanism | lErasure Coding : Support N+M redundancy protection, M supports 2, 3 or 4, suitable for SSD or HDD main memory lMultiple copies: 3 copies, etc. | ||
| storage access protocol | iSCSI, SCSI, OpenStack Cinder | ||
| System Security Policy | Disk level, node level, rack level | ||
| Key Features | SmartThin, SmartDedupe & SmartCompression, SmartQoS, SmartAuditlog,SmartEncryption, HyperSnap, HyperClone, HyperMetro, HyperReplication (A), HyperReplication (S), MultiPool, DIF. | ||
| data self-healing | Automatic parallel reconstruction with efficiency up to 4TB/hour | ||
| Deployment method | Separate computing and storage deployment, computing and storage fusion deployment | ||
| Compatible Platforms | HUAWEI FusionSphere , VMware vSphere, MicroSoft Windows Hyper-V, OpenStack and containers④,etc. | ||
| Chassis size (H×W×D) | lKunpeng model: 86.1mm×447mm×790mm lx86 models: 3.5-inch hard drive chassis: 86.1mm×447 mm×748mm 2.5-inch hard drive chassis: 86.1mm×447 mm×708mm | lKunpeng model: 175mm×447mm×790mm lx86 model: 175mm×447mm×748mm | lKunpeng model: 86.1mm×447mm×790mm lx86 models: 3.5-inch hard drive chassis: 86.1mm×447 mm×748mm 2.5-inch hard drive chassis: 86.1mm×447 mm×708mm |
| Maximum weight per chassis (including hard drives) | lKunpeng model:≤32 kg lx86 model:≤34.1 kg | ≤65kg | ≤32kg |
| Working temperature | 5℃～35℃ | 5℃～35℃ | 5℃～35℃ |
| Working environment humidity | 8% RH～90% RH, no condensation | 8% RH～90% RH, no condensation | 8% RH～90% RH, no condensation |
Note:
①NFS/CIFS/HDFS/S3 supports TCP/IP network, DPC client supports R oCE /IB network , and block storage only supports front-end business network and storage Internet using the same network;
②Multi-version feature is suitable for object storage;
③The intelligent management feature is suitable for big data storage;
④It supports iSCSI docking containers.
4008-870-872
Click Inquire
PowerLeader Server
PowerLeader ZiQiang
粤公网安备 44030902000103号
