FlashBlade Enterprise Features

NVIDIA

Audience
Public
Product
FlashBlade
Content Type
User Guides
Technology Integrations
NVIDIA
Source Type
Documentation

FlashBlade//S enables enterprise-level data management for AI at scale. Using a distributed metadata architecture, it offers multi-dimensional performance on a consolidated platform with NFS, SMB, and S3 protocol access. The cloud-based Pure1® data management platform provides a single view to monitor, analyze, and optimize your storage from anywhere. Key capabilities provided by FlashBlade//S and the Purity operating system include:

  • Multi-dimensional performance: This is the ability to deliver very high throughput and high performance to support multiple workloads simultaneously, including workloads with:

    • Any file size (small or large)

    • Sequential or random i/o

    • Batched or real-time jobs

    • Very large number of files

    • Metadata performance for file access

  • Intelligent architecture: FlashBlade was built from the ground up to truly leverage the performance and efficiencies of flash via our DirectFlash architecture. Simple to deploy, manage, and upgrade without requiring constant tuning.

  • Non-disruptive upgrades: Maintenance operations, hardware and software upgrades, performance and capacity expansions are completed without disruption.mult

  • Always available: Data is the lifeblood of your AI pipeline. FlashBlade provides enterprise-grade availability and resiliency to ensure high availability every day and without planned downtime.

  • Dynamic scalability: Seamlessly scale not only capacity but also performance, metadata, number of files and objects, and more.

  • Ready for cloud adjacency: Connect workflows from on-prem to public cloud.

  • Zero Move Tiering : This enables intelligent data access across heterogeneous all-flash clusters that optimize both performance and cost.

  • Multi-protocol support: FlashBlade supports S3, NFS over RDMA and SMB.

  • Reliability: This solution keeps organizations up and running with: