The Internet of Things (IoT) is all about scale. The idea is that millions and millions of little transactions will be performed on an ongoing basis. If done correctly, the way just about everything is done will be transformed. Then again, if the telecom and IT industries drop the ball, the IoT will melt down before it truly establishes itself.
That basic concept means two things: Each of these transactions must be done very inexpensively and fluidly, and new ways must be found to sort through the mountains of data that is collected.
One way of cutting costs is driving integration. Generally, costs go down and flexibility increases as chips aimed at specific tasks are combined and concentrated. At the GigaOm Structure 2014 this week in San Francisco, Diane Bryant, Intel’s Data Center Group’s general manager, said that it is working on a project that will combine Xeon processors with field programmable gate arrays (FPGAs). The goal, Rachel King reports, is a platform that is highly customizable.
A discussion at the conference between Actifio CEO Ash Ashutosh and DataGravity CEO and Co-founder Paula Long explored what this deluge of data will mean. The bottom line is that the concept of storage, including how and what is stored and how that data is named, will change dramatically:
Long said that this change will require the already high skill bar for IT workers to keep growing in order to answer key questions in the coming data deluge: How does information flow in the organization, what new data governance challenges do we have to deal with and how much data do we keep? And those IT workers will have to alter the network too, she said. “Networking has to completely change. Networks will have to start worrying about application data not just packets,” Lon said. “You won’t be able to aggregate and parse small sensor packets fast enough otherwise and IoT is a real time thing.”
Intelligent handling of data also is behind a new product from Irish startup Davra Networks. It sounds similar in nature to the approach discussed by the panelists, at least at a high level: There is so much data that the only prudent way of handling it is to figure out what is needed quickly. Whether, how and when resources are spent on a specific piece of data becomes a much more important issue:
In Davra's solution, data is gathered, filtered, and managed near its source, and only relevant information is sent to the cloud to be turned into insightful business intelligence calling for action. Following Cisco’s fog computing concept, simple sets of rules running on the gateways’ embedded computers can enable local intelligence.
Much of that slicing and dicing of data will be done by low-cost chipsets, such as the one being created by Intel. Sequans, Marvel and MediaTek also made moves aimed at the IoT.
Sequans, according to InfoTech, has introduced the Colibri LTE Platform, which consists of the SQN3221 and SQN3241 baseband and RF chips, the company’s LTE protocol stack, IMS client and software package for device management and packet routing.
MediaTek, according to eWeek, used Computex 2014 in Taiwan to introduce LinkIt, a development platform for the company’s Aster SoC, which measures 5.4 mm by 6.2 mm and is designed for wearable devices. The same story said that Marvell on June 3 introduced embedded microcontrollers that support Wi-Fi, Bluetooth and Zigbee. The SoCs feature EZ-Connect software.