Shares of storage technology vendor Nutanix (NTNX) are down 69 cents, or 4%, at $16.25, a victim of today’s general sell-off in the broader market, but yesterday the news flow was happier, as the company announced a “multi-year partnership” with International Business Machines (IBM) to sell Nutanix’s software on top of “hyper-converged” computing devices from IBM.
The products aren’t hear yet: further details are expected later this year with actual products vaguely promised before the year is out. The announcement was a statement of intent for the two companies to create something unique versus Nutanix’s original appliance offerings, and what IBM could have done on its own.
Nutanix’s head of marketing, Howard Ting, and IBM’s head of its “Power” computing operations, Stefanie Chiras, were kind enough to take a few minutes to talk with me by phone about the announcement.
For Nutanix, this is its third “OEM “ deal, after supplying both Hewlett-Packard Enterprise (HPE) and Cisco Systems (CSCO) with its storage software to run on top of those partners’ hardware. Previously, Nutanix had deals to deliver a combo of hardware and software to Dell, for example. This deal gets Nutanix a much wider platform for distribution by having the products sold by IBM’s global sales team.
For IBM, this is its first foray into so-called “converged” infrastructure, where a single machine incorporates a combination of storage, networking and compute.
“It is a big deal that this is the first non-Intel offering” in the converged marketplace, said Ting, meaning, the first device that doesn’t run on an Intel (INTC) “x86” microprocessor.
Chiras noted that the Power chips, which IBM has designed for years, have been tuned to run compute cycles on the open-source Linux operating system. The chips and the system, along with Nutanix’s software, have been designed to support things such as “cognitive” workloads, said Chiras, such as a bank, perhaps, trying to run financial models through deep-learning algorithms.
Power has been optimized for a number of things that Chiras thinks will make Linux efficient for tasks such as “big data” and “analytics,” things such as better handling of large cache sizes, a focus on greater bandwidth in the processor than is the case in Intel parts, and greater “multi-threading” of applications. She noted IBM’s “Open CAFE” technology provides for ten times the throughput of a comparable x86 part doing similar workloads.
I also asked Chiras about True North, a project that’s been going on for years in IBM’s research division, to develop chips focused on tasks such as artificial intelligence. Chiras said the Power team “work band in hand with our research team, absolutely, on things like accelerators and I/O [input/output].”