VAST Data builds full stack AI infrastructures with Cisco and Equinix

VAST Data builds full stack AI infrastructures with Cisco and Equinix

VAST Data’s unified data platform is ending up in various AI infrastructures. An expanded partnership with Cisco and a brand new deal with Equinix brings hardware options optimized for GenAI.

A VAST Data partnership with Cisco was already in place. For example, that company already uses VAST’s unified data platform for Cisco Nexus HyperFabric AI clusters. Now VAST Data Platform is being bundled with Cisco’s UCS servers as an integrated solution. As a result, VAST’s Disaggregated Shared Everything (DASE) architecture is intimately connected to Cisco Nexus HyperFlex hardware. This architecture is specifically designed to excel with parallel workloads. This is essential for GenAI deployment, as well as for the highly parallelized GPUs on which GenAI runs.

Also read: VAST further expands Data Platform: InsightEngine makes all RAG data available in real time

“Enterprises are seeking ways to modernize their data infrastructure to accommodate rigorous AI workloads, while consolidating and simplifying their technology stack to limit complexity,” said John Mao, Vice President, Technology Alliances at VAST Data. “This tight integration and joint selling motion between VAST and Cisco will help to accelerate enterprise AI adoption by providing end-to-end visibility of compute, networking, storage and data management – allowing organizations to seamlessly build and scale their AI operations.”

Tip: Rubrik and Cisco collaborate to identify cyber attacks

Equinix partnership

Another, new partnership is with data center company Equinix. With an Nvidia certificate in its pocket, VAST Data is going to implement these capabilities within Equinix’s IBX data centers. Specifically, this involves running VAST software on Nvidia DGX systems. A single system of these can already deliver a gigantic amount of GPU and CPU power, while also collectively acting as a modular supercomputer thanks to lightning-fast interconnects. Often this takes place via Nvidia DGX SuperPODs, which can interconnect thousands of GPUs. This is not uncommon: Oracle, for example, also interconnects 131,072 Blackwell chips via an OCI Supercluster.

Nvidia does not make the data link itself, so VAST Data acts as one of its much-needed partners. Only through this data component can there truly be a full stack AI infrastructure. Through Equinix, VAST Data provides end-to-end encryption on data, “nearly unlimited” snapshots, tools for data retention, auditability and protection against ransomware.

Equinix’s infrastructure also enables hybrid cloud computing along with VAST’s offerings. VAST DataSpace enables customers to use a hybrid cloud approach through CSPs with a global namespace through VAST DataSpace. Preparing data for AI deployment can also be managed using VAST. For this, the company is using new data structures with parallel file storage and object storage for optimal AI training and distribution.

Also read: Inside Equinix PA10: Paris datacenter sets green example