Why flash storage?

As GPU accelerator architectures take performance of AI, deep learning and big data analytic tasks to the next level, the accompanying storage requirement for large data sets to feed a training environment become increasingly demanding too. When it comes to an investment as significant as an NVIDIA DGX-1, you want to be sure that you’re seeing its maximum performance, and that the storage you pair it with will still be capable and scalable enough as GPU technology advances even further. All flash storage appliances are the only products that can deliver the data at a fast enough rate so not to cause a bottleneck.

30x Higher Throughput than CPU Server on Deep Learning Interface

With these challenges in mind, the engineers and data scientists within our deep learning & AI teams at Scan have configured, tested and optimised several all-flash options to deliver the performance required whatever your budget.

Tesla V100x30
Tesla P100x10
Dual CPU ServerX
Workload: ResNet-50 | CPU: 2x Xeon E5-2690 v4 @ 2.6GHz | GPU: add 1x NVIDIA® Tesla® P100 or V100 at 150W | V100 measured on pre-production hardware.
Pure Storage Logo

NetApp All-Flash Storage Arrays

Cost effective all-flash solutions based on a storage fabric using OnTap software architected to allow both scalability and flexibility - whether you need local fast storage, cloud back-up or a combination of both.

Find out more
Pure Storage Logo

Pure Storage Flashblade

Modern workloads, like deep learning and AI, are massively parallel – and designed to run on parallel computers with parallel storage for the highest performance. FlashBlade is architected from the ground-up for modern data analytics – delivering powerful cloud-era flash that’s fast, big, and simple.

Find out more

Nimble Predictive All-Flash Storage

Experience the Power of Predictive HPE Nimble All Flash Arrays combine a flash-optimized architecture with HPE InfoSight Predictive Analytics to achieve fast, reliable access to data and 99.9999% guaranteed availability.

Find out more

Introducing Intel Ruler SSD Storage

This new form factor delivers the most storage capacity for a server, with the lowest required cooling and power needs. The next-generation “ruler” form factor SSDs are due in 2018 and will use Intel 3D NAND & Optane technologies to enable up to 1PB in a 1U server.

Find out more
Find out more Find out more