With PCs waning, Intel’s got another bullet in its gun pointed at the Internet of Things: huge knowledge of moving, storing and managing enterprise data.
The market has focused on two of Intel’s obvious Achilles’ Heels: the lack of a low-power embedded mobile processor to compete with ARM, and the slow death of the PC as the centerpiece of our digital world. But as pundits (like me) grouse about Intel’s slow progress at righting a perceived listing product portfolio, it turns out the company has a cogent plan to capitalize greatly on bringing embedded to the Internet of Things (IoT). Intel will leverage its heavy resources in enterprise data.
Oh, they’ll still need the recently unveiled low-power Atom successors called Bay Trail and Quark. And there’s no time to waste in bringing the new two-in-one laptop/notebook/tablet concept to market this Christmas. But Intel has quietly been nudging other pieces around on the big board into what might be a winning strategy, including bringing the company’s deep IT enterprise experience to bear on a world that will rely on embedded to make the connections. That new world relies on infrastructure to make easy the task of moving, managing, abstracting, securing and monetizing all that connected machine-to-machine (M2M) data.
To learn firsthand about Intel’s plans for intelligent systems, I turned to Ryan Brown, director and chief of staff for Intel’s Intelligent Systems Group (ISG). At times Ryan is frustratingly vague because Intel has yet to announce concrete plans for M2M.
However, Intel’s still-nascent Intelligent Systems Framework for M2M is not only a set of connectivity recommendations, it will likely be first catalyzed by Wind River’s MQTT-based Intelligent Device Platform (IDP) first announced at IDF2012 and further codified at IDF2013 in San Francisco. As well, McAfee’s Embedded Control and Global Threat Intelligence products could possibly secure the end nodes or clusters of low-intelligence nodes controlled by local concentrators (Wind River and McAfee are both Intel companies.)
Finally, the new enterprise vision for Intel’s vPro adds the vector “productivity” to the IT value proposition—something we believe clearly emphasizes Intel’s focus on data that enters the enterprise from the Internet of Things. Although vPro currently relies on Ivy Bridge and Haswell processors running Windows, we expect Intel will roll vPro over to the Bay Trail Atom family and eventually onto Linux and Android operating systems (Figure 1). Supporting these embedded devices and platforms is the only way Intel can touch all the money-making parts of the M2M-connected Internet of Things.
(Edited excerpts follow.)
Q: We’ve been calling the transition from networked PCs over the Internet names like cloud connectivity, cloud computing, machine-to-machine, the Internet of Things, the Internet of Everything, and so on. Clearly everything is or will be connected to everything, and embedded devices will become increasingly important. What’s the official—if not evolving—view of what Intel calls “Intelligent Systems”?
A: There have been these compute model transformations going on over time. Whether it’s PCs, the Internet, mobile computing or all things that go around mobile computing, the next real transition point we think—despite what people call it (and you mentioned many of them)—is that all of these devices are becoming connected. Connected at the edge [of the cloud], connected to each other and up through some system to the cloud. We are currently using that [concept] as the umbrella under which Intel obviously has verticals. M2M is one of those verticals that’s very interesting to us from a technology and customer-need perspective. We can quote all kinds of numbers to justify how big the market could be but people are actually asking different questions such as: “Do I really want to connect this device?”
That’s a question that wouldn’t have even been on the radar before [the IoT]. So in that context, there are two pieces. One, there’s this opportunity for all of these things to get deployed, such as a new factory where the simple question is how to integrate the connectivity, security and the manageability pieces into the device. And a related question is what [technology] is needed to be able to move data into and out of the device into the bigger system?
The second piece is the existing infrastructure that many IoT advocates don’t want to talk about. That is: how do you actually get these things that are existing in the marketplace—like factories, transportation, infrastructure—connected together so they can actually communicate over LAN, 3G, Wi-Fi or other connection?
From Intel’s perspective, for both of these fundamental questions, we see a lot of opportunity for the industry to move forward rapidly by recognizing that all of this infrastructure will not be refreshed overnight. But users want to get some of the data off of the machines and use it. Ultimately the data will drive the transformation; it won’t be driven by merely bolting a new box on the side of a machine. How will the data allow me to do something differently with my business? What’s that data allowing me to do? How can I monetize it? Can I create a new service from it? How can I make different decisions?
Q: Are you saying that M2M is just one step in a fundamental shift going on in business?
A: From a big picture perspective…M2M is one of the internal-to-the-transformation vehicles that’s going to actually kick off the transformation I just mentioned, and Intel’s been paying attention to it for a long time. We’ve been placing devices with some companies to help them learn fast and for all of us to learn how to tweak [the data connectivity problem]. But let me be clear: the next transformation is going to take place because of the data. It will drive decisions on new equipment, new devices and new infrastructure but it will also drive change within existing structures. There’s a lot of value in connecting existing devices and this is what gives credence to the M2M mission of connecting devices together.
Q: Intel’s vision sounds in-phase with your partner Microsoft’s where Steve Ballmer revealed that all of Microsoft is writing code for the cloud. The company’s embedded roadmap shows various flavors of Windows Embedded and future Windows Embedded 8 products geared around data-centric verticals like point-of-sale (POS), handheld devices, the connected car and other small footprint (code-wise) devices such as smart meters or home automation gateways. (Figure 2 shows Microsoft’s mission and vision for intelligent devices.)
What is Intel doing, product-wise?
A: We drive all of our silicon products into embedded markets. For example, the latest Haswell and before that Ivy Bridge and Sandy Bridge processors were applicable to embedded. Atom, in Bay Trail and Clover Trail versions, also apply to embedded. We’re working closely with Wind River and McAfee to deliver a more complete solution to the marketplace. And although Intel’s bread and butter is silicon—we do that really, really well—we’re working with our customers to understand what they need in this space.
I think the biggest challenge is actually looking at the industry problems. How can Intel go solve this? The fragmentation and diverse uses in these [M2M] markets, meshed with new uses and legacy environments is a huge challenge. Add in global and local market needs and standardization…these are all things that Intel has historically done to help drive markets. We’ve applied horizontal solutions to try to drive out some of the vertical market complexities.
Q: Can you be more specific?
A: To start, we are using the building blocks that we have on the silicon technology side and looking to apply them to the problems at hand. For example, I could look at a vending machine and make a case to install a higher-level Ivy Bridge or Haswell type device that can actually gather data from all the machine’s sensors and do some interesting filtering/balancing analytics on that data to help the machine’s owner make better decisions. Now, aggregate that over multiple points and the machines’ owner now has a better picture of usage patterns of their vending machines which becomes really compelling.
It’s not as much about “what’s Intel driving specifically” beyond Haswell, because some of the chip’s security and manageability features come into play [in this scenario]. The real question for me is how can we use Intel compute technology to help solve these new, emerging problems.
Q: Again, can you be more specific about what Intel is doing to catalyze the vision?
A: We are definitely working on some things; some of them I can’t necessarily comment on. What we’re trying to do is bring together the technologies at the right level so we can deploy and help customers solve the problem.
Let me define more of the problem so you can get a better idea of the complexity customers face who have not concerned themselves with these challenges before.
For instance: how do you go build an embedded M2M device, besides the basic CPU chipset, comms, software, protocol stack, antenna design and then actually deploy it into an operational environment that is not a traditional IT environment? We see there’s value in bringing together some of these pieces that customers are going to need. That’s about all I can say at this time.
Like I said: the value’s in the data. So the customer’s real question is: how can I get access to the data the fastest? How can I get my pieces deployed and get the fastest time-to-money and start making better decisions? We are using the great technology building blocks Intel already has like Haswell, which brings the power envelop way down. But it’s still a bit “heavy” for non-wired applications.
But as you go down into our Atom space you get into the Bay Trail solutions. The next-generation versions will bring some interesting power/performance capabilities [Figure 3]. We’re working very closely with Wind River Systems to bring some of their software pieces in and make sure they run really well and coordinated with Intel products; these [capabilities] exist today. On top of that, we’re pulling in the security features from McAfee like their Embedded Control. The key question there is how are we locking down the device at the base layer so that the end product only runs what the owner wants it to run? Intel is making sure that all this works together when the device is deployed and only runs the pieces and parts that it’s supposed to run. Further up the stack, McAfee’s Secure Defender becomes applicable.
Let me say it again: we want to make it easy for the customer to deploy these M2M solutions, and it’s really, really important to us that this all works well together.
Q: What about Intel’s Intelligent Systems Framework and vPro for connectivity and manageability?
A: There’s more there, but nothing that I can talk about. We’re working in that space and we’re setting up some criteria for how systems should be thought of as you get into this space. We’ve been working closely with Wind River on their IDP and their M2M protocols. Intel has not been pointing people specifically to IDP because we’ll be supporting a broad range of products and vendors [besides just Wind River].
As for vPro on Atom, we continue to look at how we can scale some of that functionality down to Atom. We see that manageability is a feature set also desired at the hardware level, but there’s nothing that I can comment on right now.
Q: What software tools are available to help in this market transition? I’m thinking of Intel’s excellent but not-well-known tools to help designers wring out power savings in Windows-based platforms. Anything like this in the IoT/M2M area?
A: The one we can talk about was launched earlier this year at Embedded World in February is “Intel System Studio Integrated Software Suite,” but I’m not sure this is the kind of tool you’re referring to in the purely embedded space.
Q: Figure 4 shows a high level summary of what you’ve been describing. Any final comments?
A: We’ve been talking about how to make devices more intelligent and how to connect them with M2M control either directly or via a bolt-on device. The third piece—what are you going to start doing with all this data—is all about analytics and turning all that information into something “actionable.” Analytics touches all of these areas in the smart system of systems or smart intelligent systems. IoT and M2M spans all of these three nodes.