Nutanix has created a new Internet of Things Division, tasked with a long term research project to solve the massive challenges required to manage edge computing at scale.

At their .NEXT event in New Orleans earlier this month, Nutanix unveiled a new Internet of Things division and a new Internet of Things strategy. This continues the company’s recent emphasis on the higher levels of the stack and positions Nutanix for what it sees as a massive opportunity as computing for non-humans continues to expand.

The IoT division has embarked on a major long-term initiative, named Project Sherlock. The executive running it is Satyam Vaghani. Vaghani, who had been the co-founder and CTO of PernixData, and came to Nutanix when they acquired PernixData in 2016 had previously been Nutanix’s Vice President-Technology. When Nutanix created a new business unit around Internet of Things and AI earlier this year, Vaghani became its VP and GM.

“Nutanix has been focusing more and more on the high levels of the stack, with things like our new Era PaaS offering, and is going even higher with our long-term initiatives like the one we have around the Internet of Things,” Vaghani told ChannelBuzz. “These go far beyond our roots in virtualization and storage, and address things like containers, serverless computing, and analytics. They are aimed at a different set of consumers. More and more, you will see us offer interesting things to developers, for example.”

Project Sherlock is one of these high-level initiatives, and this one has a very long-term horizon.

“We considered where the company is going to go next, in the next five to seven years,” Vaghani said. “We agreed with many analysts that computing is moving to the edge in places like hospitals and factories, retail stores and airports. Last year, in 2017, edge devices produced 256 zettabytes of data, and that is forecast to rise to 600 zettabytes by 2020. This will be much bigger than the cloud – 40 times bigger.”

That realization, Vaghani said, was the genesis of Project Sherlock.

“We see edge computing growing massively, and the question was ‘what do we do about it, “ he stated.  “Every use of computing in the past was meant for humans to consume, but the computing of the future will be machine-oriented. We have to figure out this different technology stack and the different needs of machines. Then, just as we have done in the data centre, we can converge things to address these important problems on the edge, and create a simple and delightful system for edge and IoT operations. This is Project Sherlock.”

Vaghani said achieving these goals requires solving three separate challenges.

“First, the burden to set up an infrastructure that does analytics at scale is tremendously high,” he said. “A single driverless car generates a half a petabyte of data per day, and you have to analyze it all to train cars to be a better driver. You thus have to ingest half a PB of data and do all the analytics – before the next set of data comes in tomorrow. That’s a tremendous infrastructure burden. For oil and gas for example, it means setting up an analytics fabric that stretches across all the oil rigs across the world. So that’s the first challenge, dealing with the volume of data that will come in, with an operating infrastructure that is simple enough to be managed easily.”

The second problem is to democratize this process so that the system doesn’t require expensive data scientists to manage it.

“Analytics and machine learning are very difficult to consume,” Vaghani said. “You need data scientists to interpret the data, and there aren’t enough of them. You have to make it so that a generalist can do this. So that’s the second challenge, to  reduce the barrier to entry to create effective machine learning applications so that a regular joe can do this.”

The third problem is to make the system work across all kinds of infrastructure.

“Technology works very well inside a private cloud or a specific public cloud, but IoT is distributed across many clouds,” Vaghani said. “It’s a true hybrid cloud problem. You really need to have an operating layer to enable this massive movement of data into a single holistic system.”

Vaghani stressed the immensity of these problems.

“That’s the bad news, that the edge cloud is difficult,” he said. “It’s because of the divergence of devices ranging from cameras to CAT scanners, and the scale involved, and the concepts involved . It’s also divided between IT and OT and  developers and data scientists, so there are also too many cooks. It’s a huge problem, considering that we only recently just got good at running VDI at scale.”

Project Sherlock will work on the design of middleware to run all this.

“In the same way that IBM WebSphere and other kinds of middleware earlier emerged to crystalize a core set of services to run Web services at scale, with the Internet of Things we have to address the same kind of issues,” Vaghani said. “What is the core set of services needed to stand up any and all types of IoT applications? This middleware will also be focused on supporting the business logic. Usually, a  developer has to write code dealing with infrastructure and not business logic, but Project Sherlock is designed to change that. Developers will still write code, but it will be more about telling a computer what needs to be done, as opposed to telling it how to do what needs to be done. That’s much more high level code, for the business logic layer rather than the systems layer. You tell the system what you want to do and the system figures out the rest.”

Vaghani cited Smart Retail as an example.

“If Nutanix had a managed service that could manage all this from a central cloud console, to run a planet-scale IoT infrastructure, you could create a brand new IoT application in a matter of seconds,” he said.”While the logical starting point for this technology – if and when Nutanix is successful in developing it – is larger enterprises, Vaghani stressed that it would need to go far beyond that.

“We see eventual adoption as being a pyramid,” he said. “It will be the Global 2000 at the very beginning because of the strategic value. But If we succeed in making this middleware layer possible, it should be available to anyone. Remember the Nokia Communicator? No one used it because while it was great technology, it was hard to use. Then the iPhone came around and the world figured out a thousand new use cases. That is how the IoT will evolve. If we make the technology accessible, it will be used by everyone. And that’s why the channel will be critical to making this happen. IoT can only become universally accessible through the channel.”