OVHcloud unveils aggressive new storage strategy based on acquisitions and new NetApp partnership

In addition to broadening out and strengthening their object, block and file storage portfolios, OVHcloud also announced a new AI initiative to facilitate easier deployments of AI pilots into production.

Alain Fiocco, OVHcloud’s CTO

This week, at their eight annual OVHcloud Ecosystem Experience event, global cloud provider OVHcloud made a major enhancement to their storage strategy. They announced a significant upgrading of their object, block and file clouds, the first two coming from the acquisitions this summer of OpenIO and EXTEN respectively, and the last one coming from a brand new strategic partnership with NetApp. They also announced a new strategy around AI to make it easier for customers to overcome problems in turning AI projects into production deployments.

“All of these storage services – block, object, file – are things that we had in the past,” said Alain Fiocco, OVHcloud’s CTO. “The problem was these services were specific around what the customer was buying in terms of compute.” If for example, they were buying hosted private cloud, they would have access only to file storage

“We are going to enhance significantly the feature set of those services and make them available throughout the portfolio,” Fiocco stated. “The extensions from our acquisitions of OpenIO and EXTEN and our partnership with NETAPP enable us to come up with a much more feature-rich service.”

Fiocco said these changes to the storage portfolio were motivated by a desire to broaden the type of offerings the company provides.

At OVHcloud, we have always had very good price-performance ratio storage solutions, but they tended to be a little bit siloed based on the kind of solutions that customers were consuming from us,” he indicated. “Bare metal cloud, hosted private cloud, public cloud – each of them had somehow different storage characteristics and storage technologies. What we really want to provide now is a fully consistent storage portfolio across all these multiple clouds for customers to consume.”

That requires the ability to offer more high-performance offerings.

“You have two dimensions in storage: high performance and high capacity,” Fiocco indicated. “We’ve always been very good at capacity. We’ve always been very good at providing very aggressive price performance ratio in the high capacity space. Now we are focusing on complementing this portfolio with high performance storage because customers do need to have this multiple tier of storage for different use cases They need to have hot storage warm storage, cold storage and eventually very cold storage – archiving. And for each of those storage tiers, they need the right level of price performance ratio, which is what we have always been very good at. That is a given. We understand how to build this infrastructure.”

The acquisitions of OpenIO in July and EXTEN in August were indispensable to this broadening out.

“We decided to accelerate our roadmap to make sure that we had the full portfolio of protocol stacks and we made two acquisitions over the summer,” Fiocco said. “OpenIO brings great technology and talent to work on high performance object storage. They are high performance, with very rich APIs. Because OpenIO was on-prem before, they also developed features that are very specific to the market, which in addition to the very sophisticated S3 APIs, include  strong  control, security and access layer.”

The other acquisition, EXTEN, brings what Fiocco termed really good technology knowhow and very good IP in the area of high performance block storage.

“EXTEN has a unique NVMe technology that lets us do super high performance block,” he said. “It enables NVME over Network Fabric which lets you build a high performance storage infrastructure for block.”

Fiocco said this upgrade to their object and block capabilities provides multiple competitive advantages.

“We kill several birds with one stone here,” he said. “With object, we were behind in the market, because we while we have a very good service for high capacity object, we didn’t have it for high performance. This bridges that gap. With object, we have very aggressive price and with our competitors you pay for accessing the storage – a few cents for every time you read or touch the data, and pay for the traffic when you move it. We only charge for the storage. So once we have multiple object tiers, we will be ahead of the curve because of the business model.”

With block storage, Fiocco said the situation was different, and that the new technology would enable OVHcloud to better execute on their goal of providing high performance block storage in the most cost-effective way.

“Normally, this is done with an HCI server, but we can decouple that with this NVMe technology, which enables us to be more effective in providing block storage servers,” Fiocco noted.

The third enhancement, to the file cloud for structured data, involves a new partnership with NetApp, which is being made public at the event.

“This is brand new and will be available starting in January,” Fiocco said. “In trying to meet the market demand for enterprise class file storage, that means working with NetApp. We will soon be making the NetApp ONTAP protocol stack available to customers in the hosted private cloud as well as in the bare metal cloud space.”

In addition to the storage strategy, Fiocco also announced a shift in OVHcloud AI strategy to take advantage of what they see as a huge market opportunity in pretty much every sector and vertical.

“The harsh reality, however, is that most enterprises struggle to scale AI pilots into enterprise-wide production,” he said. “This is exactly the challenge that OVHcloud has decided to tackle. We want to make AI production grade. We want to provide an ecosystem, a portfolio of AI applications and models based on the ecosystem.”

Making AI production grade so data scientists can deploy their model without having to take care of the infrastructure underneath is the objective.

“That is what we do for a living, so people can bring their data and their algorithms and OVHcloud will manage the infrastructure underneath.

“We are not going to be the ones providing specialized vertical AI models,” Fiocco indicated. “Our partners will do that. We provide the platform for them to enable their specialized models. The customer gets the results and gets the bill – GPU used x number of seconds. That’s it – very simple. It removes the complexity of doing all the engineering and automation, and enables customers to move from POC stage to fully productized stage.”

Fiocco stressed that this AI strategy is new and not something that OVHcloud has had until now.

“It will come out in three weeks, and be available at the end of November,” he noted.

Leave a Reply

Your email address will not be published. Required fields are marked *