Skip to content
Techzine Europe
  • Home
  • Topstories
  • Topics
    • Analytics
    • Applications
    • Collaboration
    • Data Management
    • Devices
    • Devops
    • Infrastructure
    • Privacy & Compliance
    • Security
  • Insights
    • All Insights
    • Agentic AI
    • Analytics
    • Cloud ERP
    • Generative AI
    • IT in Retail
    • NIS2
    • RSAC 2025 Conference
    • Security Platforms
    • SentinelOne
  • More
    • Become a partner
    • About us
    • Contact us
    • Terms and conditions
    • Privacy Policy
  • Techzine Europe
  • Techzine Netherlands
Techzine Blogs Infrastructure Pure’s FlashBlade//EXA should solve storage bottlenecks in AI and HPC
4 min Infrastructure

Pure’s FlashBlade//EXA should solve storage bottlenecks in AI and HPC

"Storage is accelerating the pace of large-scale HPC and AI evolution."

Insight: Storage

Sander AlmekindersMarch 11, 2025 3:00 pmMarch 31, 2025
Pure’s FlashBlade//EXA should solve storage bottlenecks in AI and HPC

Traditional storage environments are not necessarily created or targeted to support modern AI and/or HPC workloads. Indeed, they can cause precious GPU cycles to be lost simply because the storage can’t keep up. Pure Storage aims to put an end to this with the announcement of FlashBlade//EXA.

With FlashBlade//EXA, Pure’s scale-out platform gets a third member, in addition to FlashBlade//S and FlashBlade//E. Pure is apparently keen to make the high performance of the new offering clear. It could have chosen to come out with a FlashBlade//X in parallel with the FlashArray naming convention (i.e. there is a FlashArray//X, besides FlashArray//S and FlashArray//E). It clearly chose otherwise. In any case, it is now clear that this version of FlashBlade is very powerful, given the rather explicit reference to exascale computing.

FlashBlade//EXA is focused on AI Factories

If we restrict ourselves to AI environments for a moment, they can be roughly divided into three categories. At the top are hyperscalers, at the bottom there’s enterprise AI. In the middle are the use cases that fall under the umbrella of the AI Factory. Examples include HPC labs and research centers, GPU clouds and highly sophisticated large enterprises. That’s the category Pure Storage is targeting with FlashBlade//EXA.

In a briefing on this news we received from Marco Bal, Principal Systems Engineer Benelux at Pure, he puts some numbers on the different categories for convenience. Looking purely at the throughput of systems in those environments, enterprise AI environments are fine with speeds of up to 1 terabytes per second. Hyperscalers, on the other hand, need more than 50 terabytes per second. AI Factories are in between those two, so between 1 and 50 terabytes per second.

Infographic met de kenmerken van FlashBlade//EXA: 10+ TB/sec. prestaties, 3,4 TB/sec. per rack, 20x bestanden in een naamruimte, 8+ jaar leiderschap in metadata, 50% kortere installatietijd.

Pure claims at launch that FlashBlade//EXA is the best-performing storage platform on the market today. It can handle 10 terabytes per second in a single namespace. Per rack, this equates to 3.4 terabytes per second. Now, we don’t have all the performance figures of all vendors handy, and Pure will most likely also have chosen the most favorable outcomes, but 10 TB/s is definitely a rather impressive performance. Mind you, these are initial test results. However, Pure expects to continue to live up to them.

FlashBlade//EXA is disaggregated and multi-modal

Looking briefly under the hood of Pure’s new FlashBlade//EXA platform, we see a modern storage architecture there. Today, that means, first of all, that it is disaggregated. That is, metadata and “real” data can communicate with the compute department of an AI stack separately from each other. Metadata and data can then also scale separately from each other. In theory, this should provide virtually infinite scalability.

FlashBlade EXA-architectuurdiagram op hoog niveau met de verbindingen tussen het rekencluster, de metadatakern en datanodes, met routes voor metadata en data via NFSv4.1 via TCP en RDMA.

FlashBlade//EXA is in fact the so-called Metadata Core of the architecture. Pure has fully optimized FlashBlade to handle metadata massively parallelized. The other storage component, the Data Node, where the actual data resides, consists of standard off-the-shelf nodes. Bal does add during our conversation that after general availability of FlashBlade//EXA, Pure plans to replace these with Pure’s Hyperscale Data Nodes, DFMs (Direct-Flash Modules) of 75 or 150 TB.

Infographic met de titel 'Metadata Core Architecture' met componenten als een log-gestructureerde metadata-engine, direct flash en gedistribueerde architectuur voor prestaties en veerkracht.

Besides disaggregated, it is also important to note that FlashBlade//EXA is multi-modal and multi-dimensional. That means that it is suitable for all types of data, where it doesn’t matter if the files are small or large, or it has to handle random or sequential data. This is important because different types of AI use different types and formats.

All of this, combined with the Purity operating system that also obviously runs on this, should make possible the claim from the subhead at the top of this story, made by Pure-CTO Rob Lee. Of course, whether storage will actually set the pace for AI and HPC innovation remains to be seen. There will undoubtedly be some marketing in that statement as well. In itself, that doesn’t really matter. As long as FlashBlade//EXA can keep up with the GPUs and the network, that means that storage is no longer an inhibitor. If Pure can deliver on that, it has already succeeded.

FlashBlade//EXA will most likely become generally available in the summer of 2025.

Also read: R&D is the lifeblood of Pure: a peek behind the curtains

Tags:

AI / AI Factory / disaggregated storage / FlashBlade//EXA / HPC / metadata / Pure Storage

"*" indicates required fields

Stay tuned, subscribe!

Nieuwsbrieven*
This field is for validation purposes and should be left unchanged.

Related

Pure Storage promises ‘one-click’ AI deployment with GenAI Pod

AI hyperscaler CoreWeave partners with Pure Storage

Pure Storage unchains data with range of new solutions

Rubrik and Pure Storage together bring three-layer cyber resilience

Editor picks

OpenAI and Google sign deal for cloud computing capacity

OpenAI and Google are reportedly set to sign a deal whereby Google wi...

Cisco throws next punch in battle for world’s smartest switches

New Catalyst C9350 and C9610 expand the battlefield

Lemon predicts sour faces over AI coding tools

Developer-skills and talent company Lemon thinks it has a handle on w...

Goodbye AIOps, hello AgenticOps: what is it and what can you do with it?

People and machines come together in AI Canvas

Insight: Storage

It’s World Backup Day, but backups alone are not enough

It started as a Reddit post in 2011: World Backup Day. Every March 31...

Rubrik expands to AWS, Oracle Cloud Infrastructure and OpenShift

From now on, Rubrik is expanding and debuting its protection and back...

Veeam focuses on migration and data freedom through hypervisor updates

Veeam focuses on migration and data freedom through hypervisor updates

Now including Oracle and Proxmox

NetApp refreshes high-end AFF A-Series: all-flash for AI

NetApp refreshes high-end AFF A-Series: all-flash for AI

NetApp's all-flash portfolio is doing very well. Last year it grew by...

Read more on Infrastructure

IBM builds first fault-tolerant quantum computer Starling

IBM builds first fault-tolerant quantum computer Starling

IBM has presented a roadmap to the first large-scale, fault-tolerant quantum computer. The IBM Quantum Starli...

Berry Zwets 11 hours ago
OpenAI and Google sign deal for cloud computing capacity

OpenAI and Google sign deal for cloud computing capacity

OpenAI and Google are reportedly set to sign a deal whereby Google will provide cloud services to the ChatGPT...

Berry Zwets 1 day ago
Nvidia and HPE are building the Blue Lion supercomputer in Germany

Nvidia and HPE are building the Blue Lion supercomputer in Germany

Nvidia and Hewlett Packard Enterprise are collaborating with the Leibniz Supercomputing Centre to build a new...

Berry Zwets 1 day ago
Goodbye AIOps, hello AgenticOps: what is it and what can you do with it?
Top story

Goodbye AIOps, hello AgenticOps: what is it and what can you do with it?

People and machines come together in AI Canvas

Sander Almekinders 35 minutes ago

Whitepapers

How to choose the right Enterprise Linux platform?

How to choose the right Enterprise Linux platform?

"A Buyer's Guide to Enterprise Linux" comprehensively analyzes the mo...

Try the latest high-end Synology backup system for free

Try the latest high-end Synology backup system for free

How do you ensure that your data is secure and can be quickly restore...

Enhance your data protection strategy for 2025

The Data Protection Guide 2025 explores the essential strategies and...

Strengthen your cybersecurity with DNS best practices

The white paper "DNS Best Practices" by Infoblox presents essential g...

Tech calendar

Kaseya DattoCon Europe

June 17, 2025 Dublin

Nutanix Cloud Day Nederland 2025

June 17, 2025 Zeist

Akamai Customer Day Benelux

June 18, 2025 Nieuwegein

Nürnberg Digital Festival 2025

June 30, 2025 Nürnberg

GITEX DIGI_HEALTH 5.0 - Thailand

September 10, 2025 BITEC Bangkok, Thailand

IT Arena

September 26, 2025 Lviv, Ukraine

Techzine Global

Techzine focusses on IT professionals and business decision makers by publishing the latest IT news and background stories. The goal is to help IT professionals get acquainted with new innovative products and services, but also to offer in-depth information to help them understand products and services better.

Follow us

Twitter
LinkedIn
YouTube

© 2025 Dolphin Publications B.V.
All rights reserved.

Techzine Service

  • Become a partner
  • Advertising
  • About Us
  • Contact
  • Terms & Conditions
  • Privacy Statement