Juniper Networks is opening Ops4AI Lab, a test lab for solutions from different vendors in the AI data center.
The test zone is important for companies that want to run AI in their own data center. Within these companies, the puzzle will need to be solved regarding which network infrastructure can best support this. In the lab, companies can test switching, routing, storage, and compute solutions from various vendors. Capabilities include offerings from AMD, Broadcom, Intel, Nvidia and Juniper.
Simpler through validated designs
Juniper Networks is additionally marketing the Juniper Validated Designs (JVDs), which should lead to accelerated time-to-value when deploying AI clusters. The validated designs are built on platforms such as Nvidia A100 and H100 compute, supplemented by storage capabilities from Juniper’s partners.
Furthermore, the company is making modifications to software where AI workloads over Ethernet should benefit from improved performance. The “Ops4AI software,” as the enhanced software is called, focuses on three things: automatically choosing the best control settings, better network utilization and end-to-end visibility from network to SmartNICs.
AI in your local data center
When running AI workloads via the cloud, it uses data center infrastructures built by the cloud provider. Juniper Networks, with the Ops4AI initiatives it is now announcing, is trying to make it more attractive to run AI locally in one’s own data center. That should be done by shortening the time required to devise and build out its local supporting infrastructure.
Also read: HPE and Juniper take aim at Cisco with network fabric