HPE Parallel File System Storage
Are you frustrated with the architectural and economical limitations of your current high-performance file storage?
HPE Parallel File System Storage is cost-effective, parallel storage for your high-performance simulation, AI, and data analytics environments running on HPE Apollo systems.
HPE Parallel File System Storage provides multiples of performance and namespace scalability, as compared to standard scale-out NAS storage, to increase the utilization of your compute nodes by removing I/O bottlenecks while enabling cost savings through storage island consolidation in a unified, high-performance namespace.
- Server configuration for NFS/CIFS/S3 access to PFSS , file systems on Protocol Servers.
- Selectable performance levels for Protocol Servers ranging from less than 2000 up to 15000 concurent users.
- Lower cost network NICs for attaching to 10 and 40 Gb ethernet networks.
The Storage Performance Your HPE Apollo Systems Need
HPE Parallel File System Storage is built on industry-standard server technology utilizing the AMD EPYC 7000 series processor family and HPE SmartMemory up to 3200 MT/s DDR4 memory.
Fast PCIe 4.0/NVMe SSDs are the standard medium to deliver performance with the ability to add HDD storage pools in the same namespace to provide cost-effective storage capacity.
High speed, low latency networks like InfiniBand EDR/HDR or 100/200 Gigabit Ethernet deliver the data to the CPU or GPU compute nodes.
NVIDIA® Magnum IO support (GPUDirect Storage) for accelerated input/output (IO) for NVIDIA GPUs, which provides a path for data to bypass CPUs.
Highly available write caching (HAWC) leverages non-volatile storage in the compute nodes for accelerated IO for write-intensive applications.
The Scalability Your HPE Apollo Systems Need
HPE Parallel File System Storage supports up to 128 storage servers in a single file system/name space far beyond the limitations of NAS storage.
Start with as little as four rack units and grow to petabytes (PBs) of capacity.
Very granular scale-out ,in increments of one rack unit.
Scale-out linearly within a namespace due to the parallel file system architecture.
Scale-out PCIe 4.0/NVMe flash performance independently from cost effective HDD capacity in the same file system.
The Enterprise Storage Features You Need
HPE Parallel File System Storage offers a comprehensive set of system availability features that provide enterprise-grade "five-nines" availability levels with multi-parity RAID 4+2, 4+3, 8+2,8+3, 3, and 4 way replication, automatic fail-over and intelligent fail-back for mission-critical use cases.
Comprehensive set of data protection features like snapshots, synchronous and asynchronous replication, file-level encryption, and asynchronous error diagnosis while affected input/output (I/O) operations continue.
Its file system supports interfaces for file (POSIX, NFS, CIFS), object (S3, SWIFT), and Hadoop Distributed File System (HDFS) for broad data accessibility across the organization.
Provides data integrity checking on data at rest and in flight, delivering higher reliability and operational robustness.
Support for auditing capability to monitor file access to a defined set of data. The resulting audit logs are stored in an immutable file set to support the compliance requirements of your industry.
The Cost-Effectiveness Your HPC and AI Budget Needs
HPE Parallel File System Storage enables "storage islands" consolidation for your HPC, AI, and HPDA environments, delivering high performance for all file sizes/IO patterns, superior scalability, and broad protocol support.
Its software-defined architecture combines a leading enterprise parallel file system, IBM Spectrum Scale, with one of the leading rack-optimized industry standard servers, HPE ProLiant DL rack servers, as storage nodes.
Focus on exploiting the different strengths of storage media (SSD and HDD) in a cost-effective way within the same file system/name space without incurring their weaknesses, delivering the performance and capacity you need in a cost-effective way.
Quality of Service (QoS) management allows to specify limits on throughput to limit the resources consumed by any fileset in order to control resource consumption in the shared storage pool. QOS management allows limits on input/output operations per second (IOPS), maximum MBs, or both.
AMD is a trademark of Advanced Micro Devices, Inc. Linux is the registered trademark of Linus Torvalds in the U.S. and other countries. Windows and Windows Server are either registered trademarks or trademarks of Microsoft Corporation in the United States and/or other countries. NVIDIA and GPUDirect are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries. Red Hat is a registered trademark of Red Hat, Inc. in the United States and other countries. All third-party marks are property of their respective owners.
Find what you are looking for?