3000 family

SFX3000 Storage Processor
CSD 3000 NVMe Computational Storage Drive
CSD300 Computational Storage Processor
CSware Computational Storage Optimized Software
ScaleFlux Announces Next Generation Portfolio - ScaleFlux CSD 3000 Series! Bringing Highly Efficient Computational Storage to the Masses!

Helping you…

Balance I/O & Compute
Reduce Data Movement
Accelerate Applications

Products that…

Parallelize Processing
Bring “Compute to Data”
Scale Low-latency Storage

CSD 2000 Drives

Quadruple Capacity
Double Performance
Halve Flash Costs
CSD 2000 Series is the only PCIe SSD with penalty-free compression and decompression… Stop compromising between Low Cost and High Performance, get both!
What is ?

A new class of storage drive

that brings compute right to the data.

State of the Art
Compute Engines

Inline compression/decompression, encryption/decryption, filtering, and more
The Latest 3D NAND
Flash Memory

4TB to 16TB Physical Capacity Drives
Customer Proven

5x-10x Better value than commodity SSDs
Learn More >

Introduction to Computational Storage

ScaleFlux 3000 Family Introduction

Watch the video

Why Choose ScaleFlux

Experienced Storage & Data Center Team
We have built a team of industry veterans with extensive knowledge of Flash, Storage, Data Center Applications. We understand the challenges Database Admins, Infrastructure Architects, DevOps & Procurement teams face in meeting the demands of the data-driven enterprise. .

Get more done

Using ScaleFlux CSDs results in higher work output per server.

Read More

Reduce your Costs of Scaling

ScaleFlux drives provide penalty-free compression, reducing the costs and increasing the capacity of your low-latency tier of Flash storage.

Read More

Break the Bottlenecks

ScaleFlux CSDs bring compute to the data to alleviate I/O, Memory, CPU and Network bottlenecks.

Read More

Easily Deploy in Standard Servers

ScaleFlux drives are available in standard 2.5” U.2 SSD and Low-profile PCIe Add-in Card form factors.

Read More

Optimize Settings for your Use Case

ScaleFlux lets users adjust the drive settings to optimize performance, capacity, application latency, and cost. No more fixed settings from the factory!

Read More

Update In Place

The ScaleFlux CSD implementation allows users to deploy new features and new programs via Firmware updates and Software downloads.

Read More

Customer Success Stories

ScaleFlux is in production with its 2nd generation of Computational Storage Drives. Our products have passed not only our own, extensive qualification test suite, but also qualification testing at dozens of customers.


Problems we hear from our enterprise customers…

“My data is growing so rapidly, scaling my infrastructure is a challenge”

“I’ve tried using compression to save costs, but I can’t afford the latency penalty”

“I need higher capacity of fast data for my business intelligence and analytics, but SSDs are expensive”


ScaleFlux CSDs with integrated compute functions to help efficiently scale performance and capacity.

Read More


Performance Gains
10x better latency
Up to 10x performance density

Over 2x storage capacity per $
Up to 10x better cost/transaction

Infrastructure Simplicity
Up to 4x smaller storage footprint
Up to 2x fewer nodes

Where can Computational Storage help?

Latest News & Blogs

Storage Industry Predictions 2022

Dr. Hao Zhong, ScaleFlux co-founder and CEO's predictions on computational storage have been included in Storage Newsletter.

Market opportunities, security challenges and the edge -- 5G predictions for 2022

ScaleFlux Co-founder and CEO Dr. Hao Zhong's predictions on 5G are included in BetaNews.

Supermicro and ScaleFlux Collaborate on Server Storage Solutions

A detailed piece in Hosting Journalist about our partnership with SuperMicro on FatTwin, a high-density 4U multi-node system -- delivers a fully integrated high performance, cost-effective platform that is uniquely suited to Redis on Flash (RoF) deployments.

Storage 2022: Active archiving, ML-enabled volumes on the rise

ScaleFlux CEO, Dr. Hao Zhong’s 2022 predictions were included in Chris Preimesberger’s ZDNet article.