Are your HPC and AI compute workloads bottlenecked by slow storage performance? WekaIO Matrix is a high performance, scalable, and parallel file system that is ideal for AI, technical computing, and mixed workloads. The flash-native, highly resilient, POSIX file system delivers the high IOPS and low-latency throughput needed for demanding compute requirements. WekaIO Matrix provides integrated policy-based tiering, so data can span from NVMe flash to object storage in a single namespace for easy management and cost-effective economics. Native support for the industry-standard S3 interface allows integration with both on-premises and cloud environments. Your organization made a significant investment in the compute infrastructure to support your analytics workloads, not allow data accessibility be the bottleneck to the overall productivity of your solution. WekaIO Matrix delivers the performance you need, so your data analysis pipelines will never be stalled waiting for data.
Reduced Time to Results Due to High Performance and Microsecond Latency
Designed for flash and NVMe technology, the WekaIO Matrix file system accesses the underlying flash media directly in its native 4 KB format so that both small files and large data sets can be processed in less time.
Low-latency InfiniBand or Ethernet support means that data locality is no longer an issue. Data is dynamically distributed for parallel access, resilience, and consistent performance.
Delivers over 6.5 GB per second performance to each InfiniBand EDR-connected client compute node, providing file system performance faster than using storage devices located directly within the compute node itself.
Reduced CAPEX and OPEX Through Improved Resource Utilization
WekaIO Matrix can leverage a single infrastructure to work on-premises or in the cloud (public or private) and elastically scale to match the requirements of your GPU or CPU resources.
Full POSIX compliance means applications won't need to be re-written for improved performance.
This architecture allows shared POSIX-based access to your data without the requirement for unnecessary copies for data protection, providing full storage resource utilization.
Supports multi-protocols: POSIX, NFS, SMB, HDFS, and S3, allowing one platform to serve your data lifecycle needs.
Cloud Scale and Economics for Increased ROI
WekaIO Matrix offers native S3 storage gateway which enables data tiering to any S3-compatible object store for cost-effective storage at massive scale.
Innovative snap-to-S3 feature enables cloud bursting, remote backup to the cloud, and cloud-based disaster recovery.
Supports private or hybrid cloud environments without modification.
With WekaIO Matrix, you can create a flexible infrastructure tuned for Hewlett Packard Enterprise hardware that can handle peak workloads – but which also scales for performance and capacity growth as system requirements evolve.
Design Adapts Easily to Changing Application Needs
WekaIO Matrix supports small and large files, random and sequential access, and accelerates metadata access with distributed parallel access.
The native flash implementation and support for high-performance interconnect fabrics helps eliminate the bottlenecks associated with legacy storage approaches, and the solution can be easily monitored, tuned and managed using an intuitive browser-based administration console.
Supports Ethernet and InfiniBand to 100 Gb and beyond, allowing full use of existing network, while accommodating future upgrades for improved performance and low latency. The software-defined platform enables flexible adaptation to future generations of infrastructure.
- HPE Apollo 2000 Gen 10 Systems
- HPE ProLiant DL360 Gen10 Server
- HPE Scalable Object Storage with Scality RING
- WekaIO Matrix for AI and Analytics Data Sheet
- WekaIO Matrix Data Sheet
- HPE Reference Configuration for HPE AI Data Node