Download the authoritative guide: Cloud Computing 2019: Using the Cloud for Competitive Advantage
Despite all the bluster about $11 trillion Internet of Things opportunity, most companies are still struggling to get even $1 of value from IoT.
As software developer Alan Cohen argues, “When it comes to the Internet of Things, it's both particularly easy to get something magical working a little bit, and particularly difficult to get the thing completed.”
Proofs of concept, in other words, come cheap in IoT land. But overcoming cost, security, and data integration problems resolved to run significant systems at scale? Good luck.
Maybe luck isn’t really what’s needed, however. Instead, what we may need is a good cloud. Though cloud may seem a counterintuitive answer to an industry trend dominated by disparate sensors and devices, it’s exactly the right approach, as AWS vice president of Mobile and IoT, Marco Argenti, told me. IoT, he says, is not really about “turning lights on or off - it’s about transforming companies, and for that you need a rich platform.”
In short, you need a home for all the data thrown off by “things.”
Seaside Yet Still Docked
Given how much attention has been paid to IoT over the last few years, it seems like we should be farther along than we are. No, it hasn’t helped that IoT has been a morass of conflicting “standards,” with hundreds of discordant alternatives each vying for supremacy. None succeeded. Oh, sure, we saw hordes of vendors joining this or that industry foundation, but none ended up as the clear, de facto standard.
In the absence of IoT standards, we’re left with a cacophony of devices throwing off data in diverse formats. A data scientist already spends the majority of her time cleansing data to make it useful for machines to digest, but with IoT this problem is compounded by the messiness of the data.
Furthermore, the technology used to gather, process, and distribute data has also remained fractured. Enterprises really don’t want to have to dig down into the weeds of IoT technologies, but that’s exactly what they’re forced to do today. A Google might have the engineering horsepower to pull this off, but most mainstream enterprises simply don’t.
If this sounds chaotic, it’s because it is. It’s also a perfect opportunity for a platform company like AWS.
AWS IoT and ‘Glamorous Glue’
Talking to Argenti, I asked him what makes AWS different. “AWS itself,” he answered. When I pressed for more detail, he told me: “IoT workloads are complex and require security at scale. They need to aggregate data from disparate places.”
As such, he continued, “You need a rich platform to be successful with IoT.” Given that IoT, like all modern workloads, is really about the data, AWS offers the ability to “Collect large amounts of data, process it at scale, understand what’s going on, and act accordingly.”
Or as Argenti’s colleague, Amazon CTO Werner Vogels, recently declared at AWS Summit in San Francisco, “What makes our businesses unique is the data we have, and the quality of that data.”
One of the key things that AWS has done well, generally, and which makes it a dominant player in IoT, specifically, is its expertise in data. While companies like Oracle have largely tried to push developers into a one-size-fits-all database approach, AWS follows a different credo: Give developers the right tools for the right job. AWS’ intent, in other words, is to deliver purpose-built engines, with each engine fulfilling its role particularly well.
This has meant that AWS, unique among the cloud vendors, has built out a bevy of different databases. AWS offers twice as many relational database services as Microsoft Azure, and nearly triple that of Oracle.
For RDBMS needs, developers are looking for fast performance, referential integrity, and don’t want the complexities of high availability. By contrast, for apps that need NoSQL, developers want infinite scale while ensuring predictable performance. Given that any given IoT app will likely need different database engines, AWS has worked hard to ensure the widest possible array of options.
AWS IoT and Data Magic
Of course, enterprises must first get their data to AWS in order for the “data magic” to happen, but Argenti insists that AWS is able to move and process large amounts of data with sub-millisecond latency.
“Historically,” he says, “the first problem we tried to solve with AWS was trying to extract data from a variety of devices. One of our advantages is that we have a very efficient communication channel with the device, e.g., MQTT, which gives a binary, optimized way to transmit as little data as needed, all while securing the channel.
As for the devices themselves, AWS can not only pull data from a wide array of sensors and other devices, but also hired the maintainer of FreeRTOS so that it could build out its own distribution, Amazon FreeRTOS. This lightweight operating system makes “small, low-power edge devices easy to program, deploy, secure, connect, and manage.”
To drive machine learning and general data intelligence closer to edge devices, AWS offers the IoT platform Greengrass. Greengrass basically takes the AWS cloud platform out to edge devices so “they can act locally on the data they generate, while still using the cloud for management, analytics, and durable storage.”
Hence, while Microsoft has trumpeted its $5 billion investment in IoT, AWS keeps building on the cloud services most developers already know and love. In this way, AWS, more than any other company, keeps lowering the bar to IoT success.
Doing the IoT Heavy Lifting
And let’s be clear: that bar needs to keep coming down, even for AWS. As Argenti put it to me, “For years IoT has been aspirational and required super skills.” Consultants would dangle the potential for trillion-dollar benefits derived from IoT, but most companies lacked (and still lack) the “super skills” necessary to making IoT work.
When I pushed Argenti on whether AWS does much to improve on this gap between desire and competence in the industry, he had two responses. The first was to call out the value of the platform: “A lot of the heavy lifting is done by us. The job is not done, but we do make it easier. We’re trying to break the digital divide.” This resonates with the AWS Summit theme of AWS taking on “the undifferentiated heavy lifting” that most enterprises are forced to slog through for any kind of application. For IoT, it’s even worse.
Argenti’s second answer was to point to the kinds of customers that are seeing success with the AWS IoT platform: Amway (connecting Amway devices to the cloud); iRobot (connecting devices to AWS cloud services); Royale International (managing a courier service to 7-11 stores); Under Armour (powering the Connected Fitness app); ENEL (energy management); and more. These aren’t name-brand Silicon Valley companies filled with uber-geeks. They’re far more mainstream.
The trick for AWS has been to constantly lower the bar to IoT success. As just one example, with AWS Greengrass you can deploy a Lambda function written in NodeJS or Python, web languages that many developers will know, making Greengrass that much more accessible to mainstream developers. In so doing, AWS has enabled “mere mortals to really transform their operations without necessarily having to hire hyper-specialized staff,” Argenti says.
Nor is the company anywhere near complete. “There is lots of opportunity for us to continue down this path,” he says. “We’re trying to democratize IoT for those that may be great at operations but need help with the infrastructure for IoT.”
It’s exactly the type of order we’ve needed in the chaos spawned by IoT’s early days. Undoubtedly we’ll see Microsoft, Google, Apple, and other big companies compete for IoT developer hearts and minds, but as with cloud generally, it’s a tough slog for anyone trying to catch up to Amazon’s developer prowess.
Matt Asay is Head of Developer Ecosystem at Adobe