Nvidia scoops up object storage startup SwiftStack
Nvidia ideas to acquire item storage vendor SwiftStack to assist its consumers accelerate their artificial intelligence, high-overall performance computing, and knowledge analytics workloads.
The Santa Clara-based mostly GPU vendor will not sell SwiftStack software package to new potential customers but will use SwiftStack’s 1space as aspect of its internal artificial intelligence (AI) stack. It will also enable consumers to use the SwiftStack software package as aspect of their AI stacks, in accordance to Nvidia’s head of organization computing, Manuvir Das.
SwiftStack and Nvidia disclosed the acquisition right now. They did not expose the obtain price tag, but they said it expects the offer to close with months.
Nvidia previously labored with SwiftStack
Nvidia labored with San Francisco-based mostly SwiftStack for additional than eighteen months on tackling the knowledge worries associated with functioning AI applications at a large scale. Nvidia identified 1space particularly practical. SwiftStack introduced 1space in 2018 to accelerate knowledge access across public and non-public clouds by a solitary item namespace.
“Basically set, it can be a way of placing the proper knowledge in the proper position at the proper time, so that when the GPU is active, the knowledge can be despatched to it immediately,” said Das, who will oversee the integration of the SwiftStack as very well long run technological innovation ideas.
Das said Nvidia consumers would be ready to use organization storage from any vendor. The SwiftStack 1space technological innovation will sort the “storage orchestration layer” that sits in between the compute and the storage to thoroughly position the knowledge so the AI stack runs optimally, Das said
“We are not a storage vendor. We do not intend to be a storage vendor. We are not in the business of providing storage in any sort,” Das said. “We perform incredibly intently with our storage companions. This acquisition is built to even further the integration in between different storage systems and the perform we do for AI.”
Nvidia companions with storage suppliers this kind of as Pure Storage, NetApp, Dell EMC and IBM. The storage suppliers combine Nvidia GPUs into their arrays or sell the GPUs along with their storage in reference architectures.
Nvidia captivated to open resource tech
Das said Nvidia identified SwiftStack desirable due to the fact its software package is based mostly on open resource technological innovation. SwiftStack’s eponymous item-and file-based mostly storage and knowledge management software package is rooted in open resource OpenStack Swift. Das said Nvidia ideas to carry on to perform with the SwiftStack crew to advance and optimize the technological innovation and make it readily available by open resource avenues.
“The SwiftStack crew is aspect of Nvidia now,” he said. “They’re super talented. So, the innovation will carry on to happen, and all that innovation will be upstreamed into the open resource SwiftStack. It will be readily available to anybody.”
SwiftStack laid off an undisclosed selection of product sales and marketing employees in late 2019, but held the engineering and assistance crew intact in accordance to president Joe Arnold. He attributed the layoffs to a change in product sales target from basic backup and archiving to AI, equipment mastering (ML) and knowledge analytics use scenarios.
The SwiftStack seven. software package update that emerged late very last 12 months took intention at analytics HPC, AI and ML use scenarios, this kind of as autonomous vehicle applications that feed knowledge to GPU-based mostly servers. SwiftStack said at the time that it had labored with consumers to structure clusters that could scale to handle multiple petabytes of knowledge and assistance throughput in extra of 100 GB for every next.
Das said Nvidia has been using SwiftStack’s item storage technological innovation as very well as 1space. He said Nvidia’s internal perform on knowledge science and AI applications had immediately confirmed the firm that accelerating the pc shifts the bottleneck somewhere else, to the storage. That played a aspect in Nvidia’s acquisition of SwiftStack, he famous.
“We acknowledged a long time in the past that the way to assist the consumers is not just to deliver them a GPU or a library but to assist them make the overall stack, all the way from the GPU up to the applications them selves. If you search at Nvidia now, we commit most of our power on the software package for different varieties of AI applications,” Das said.
He said Nvidia would completely assistance SwiftStack’s buyer foundation. SwiftStack statements it has all around a hundred twenty five consumers. It merchandise lineup incorporated SwiftStack’s item storage software package, ProxyFS file procedure for built-in file and item API access, and 1space. SwiftStack’s software package is built to operate on commodity hardware on premises, and its 1space technological innovation can operate in the public cloud.
SwiftStack put in additional than eight a long time increasing its software’s capabilities since the company’s 2011 founding. Das said Nvidia has no rationale to sell the SwiftStack’s proprietary software package due to the fact it does not contend head to head with other item storage companies.
“Our philosophy right here at Nvidia is we are not attempting to contend with infrastructure suppliers by providing some sort of a stack that competes with other peoples’ stacks,” Das said. “Our purpose is just to make people today successful with AI. We imagine, if that takes place, most people wins, which includes Nvidia, due to the fact we believe GPUs are the finest platform for AI.”