VMware Inc. is boosting support for one of the latest bare-metal instances on the Amazon Web Services Inc. platform.

On Tuesday, VMware unveiled an enhanced version of VMware Cloud on AWS that will facilitate the running of specific instances on the AWS cloud to realize economies of scale and provide more options for additional storage.

“We have an announcement to make on some of the hard work that we’ve been tackling to improve the flexibility of customers to move from on-premises to the cloud with better sizing options … with m7i.metal-24xl,” said Fred Wurden (pictured, left), vice president of commercial software services at AWS. “We can size storage, and data is so critical in terms of really understanding how that works with new workloads that we can run on this particular instance. It will be available in quite a few regions shortly.”

Wurden spoke with theCUBE industry analyst John Furrier at the “Supercloud 5: The Battle for AI Supremacy” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. He was joined by Narayan Bharadwaj (right), vice president and general manager of cloud solutions at VMware, and they discussed how the latest news will improve workload management and migration.

Decoupling compute and storage

The implementation for VMware Cloud on AWS comes with an attractive storage option for enterprise users, according to Bharadwaj. Customers can manage storage hardware attached to instances using VMware’s Cloud Flex Storage service, decoupling it from compute.

“Customers told us it would be nice to dial up compute and storage independently to get an even better total cost of ownership when you’re running in the public cloud with VMware Cloud and AWS,” Bharadwaj said. “So, we’ve done exactly that; this is a complete decoupling of storage and compute. That decoupling really gives customers a better cost model, enables customers to migrate workloads to the public cloud faster, cheaper.”

The latest announcement from VMware and AWS can tailor performance to the workload, an important element in both speed and efficiency for migrations.

“You can have acceleration, but that’s only one portion of the equation,” Wurden said. “It usually comes back to a couple of things, both your network performance but also your storage. You have to have the system performance, whether it’s to Elastic Block Store or your network, performance [is] completely understood and modeled quickly and efficiently for your particular workload. This is one example of being able to do that.”

Use of AWS m7i.metal-24xl instances is based on a custom CPU from Intel Corp.’s most recent Sapphire Rapids server chip series, with 96 vCPUs and 386 gigabytes of memory. This can help enterprises with AI and machine learning applications.

“What I like about this m7i architecture that we have is the fourth-generation Intel AMX, Advanced Matrix Extensions,” Bharadwaj explained. “It creates what I call pragmatic AI/ML in the market where customers can run their exiting enterprise workloads and their AI/ML workloads side by side on the same cluster. Whether you are doing any kind of data pipelining or inferencing or testing or optimizations, you can really take advantage of this in this new architecture. You have the best of both worlds.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the “Supercloud 5: The Battle for AI Supremacy” event:

Photo: SiliconANGLE