CAMBRIDGE, Mass. -- Data continues to grow at an exponential rate, especially within enterprises that exploit internet-of-things...
sensors and edge computing. AWS has amped up its IoT efforts, as it hopes to enter this lucrative market.
AWS has steadily increased its presence in the internet of things (IoT) since it acquired startup 2lemetry in 2015. Six IoT-focused services unveiled at re:Invent 2017 demonstrate its attention to edge computing. And in early April, several AWS features released to general availability at the AWS Summit in San Francisco apply to IoT workloads and augment AWS' cloud IoT platform.
One of those examples, AWS Greengrass ML Inference, lets IT teams run machine learning workloads on local connected devices. Machine learning and AI seem poised to converge with IoT workloads to filter out which data to collect and which to discard -- and AWS hopes to capitalize.
"[Amazon] has been pushing hard," said Frank Gillett, IoT analyst at Forrester, here at last week's MIT Enterprise Forum's Connected Things conference. "What we hear from system integrators and developers is that there's a lot of interest and comfort with AWS."
Serverless and IoT fit together
IRobot, a consumer robotics company based in Bedford, Mass., runs production applications for its connected robots on AWS. IRobot relies on roughly 30 cloud services, such as AWS IoT Core and AWS Lambda, in an entirely serverless deployment, to lower operational overhead and focus on hardware and software development.
"That's been a great enabler for us, both AWS in particular and serverless technology in general," said Ben Kehoe, cloud robotics research scientist at iRobot. "We're not burdened with maintaining servers, and we're not tasked with the scalability of the platform we're running on. All of that is coming from the cloud provider."
Kehoe said he is particularly interested in two AWS features: Lambda's support for Simple Queue Service as an event trigger and AWS' Private Certificate Authority, which can help IoT teams manage private certificates.
A Python shop, iRobot has embedded code in its products for years and likely avoided the learning barrier most companies will face with IoT integration. "We know how to put code on devices, how to update that code [and] how to secure it," Kehoe said. "We can do all those things ourselves and have mature processes for them."
Kehoe said he has no qualms with AWS' cloud IoT platform, though he said he wants better visibility to manage and track IoT deployments that integrate Amazon cloud services across many different accounts and the way in which permissions are given.
Cater to the adopter
For IT shops that lack such IoT expertise, AWS aims to lower the barrier to adoption with its cloud IoT platform and Greengrass software. Greengrass enables limited compute on edge devices, but not all enterprises have the personnel in place to realize its potential.
"When you look at who is doing IoT, they're not IoT people and they're not developers," Gillett said. "At the end of the day, you have a product manager and a business strategy owner who have to figure out how they're going to change products. They're looking for a solution they can understand, work with and grow with."
Applications, instead of platforms, would help these customers who otherwise rely on consultants and systems integrators to deploy IoT, Gillett said. "There's a build-buy question that a lot of people are going to struggle with," he said.
AI and IoT change the game for data management
Meanwhile, manufacturers and other industrial IoT buyers are more likely to turn to vendors who tailor offerings to that market, such as GE, Siemens and Schneider Electric.
Gillett said he thinks AWS could fill that application gap either through acquisition or development. But he does hear more "energy and momentum" around AWS than the Azure IoT Suite. Despite some recent IoT investments and leadership changes, the Google Cloud IoT platform is an even more distant competitor, he said. But, as with other cloud technologies, inertia factors into the decision.
"Once you already have a default [cloud provider], it takes quite a bit to move you off that default, particularly at the fast pace that these companies and this market changes," Gillett said.