Skip to content

TECH PARTNERS

Addressing AI Infrastructure Challenges with Wasabi and Hammerspace

April 11, 2024

In February, Hammerspace announced a new storage architecture for AI and GPU computing called Hyperscale NAS. This new architecture combines HPC-class file system performance with the standards-based simplicity of enterprise NAS, and gives data driven organizations a new option for AI training, GPU computing, and other forms of high-speed data processing. 

hammespace hyperscale NAS diagram

This new architecture solves the problem of keeping GPUs fully utilized when the GPU cluster is co-located with the data storage cluster, but what about organizations that want to use high speed cloud storage, and access cloud compute resources as part of their data pipeline or workflow? 

The Wasabi and Hammerspace solution 

Wasabi and Hammerspace can help solve these problems with a validated solution for hybrid-cloud storage and multi-cloud file sharing. 

diagram of hammerspace and wasabi working together

Hammerspace Data Orchestration automatically places files and objects on Wasabi Hot Cloud Storage. This joint solution has been validated and tested.

Use Wasabi cloud storage to avoid vendor lock-in and lower your total cost of cloud infrastructure. Hot Cloud Storage seamlessly integrates with compute services from hyperscale cloud providers as well as Equinix Metal, Hivelocity, IBM, Vultr, and more.

Learn more 

If you are looking for hybrid-cloud solutions for AI training and other forms of high-speed data processing, this analyst white paper talks about the need for a new class of storage architecture and digs into why solutions like Wasabi and Hammerspace are needed for the Next Data Cycle. 

Related article

data protection
TECH PARTNERSSpicy Bytes with Synology: Hot wings, hotter takes on data protection

Most Recent

CapEx to OpEx: How to calculate your true surveillance storage cost

Moving surveillance storage to the cloud was supposed to lower costs. For most teams, it just created cost unpredictability. This article breaks down what billing actually costs for surveillance workloads and how to calculate your true TCO before you commit.

Unstructured data management for AI, part 2: Storage infrastructure your AI won't outgrow

Learn how to design a cloud data lake that scales with your AI workloads, covering object storage architecture, key design decisions, governance, and what to avoid when building your infrastructure.

The AI pipeline may be your biggest security blind spot

Your AI model is secure, but is the data pipeline beneath it wide open? Learn why defense-in-depth at the storage layer is the step most AI teams are missing, but every AI team needs.

SUBSCRIBE

Storage Insights from the Storage Experts

Storage insights sent direct to your inbox.

Subscribe