Powering the Internet of Things with Real-Time Hadoop

<< back Page 2 of 3 next >>

connected jet engine, the engine can report issues to a central system from mid-air and have an alert go out in real-time to the airlines’ maintenance department that either a corrective or preventive action is needed. Proactive fault response like this is comparable to the difference between seeing the “check engine” light come on in your car, and having a mechanic show up at your home because your timing belt is cracked and your car is not safe to drive.


One place where the Internet of Things has been most visible to the general public is on the wrist, thanks to wearables like the myriad fitness trackers and smart watches on the market (and the Apple Watch that threatens to disrupt all of them). As wearable devices pack more and better sensors into their cases, their value as data collectors increases. One prime example of this new utility is in healthcare, a trillion dollar industry that is rapidly evolving. By being able to compare multiple data points (e.g. heart rate, skin temperature, insulin levels) from a user in real-time, systems will have a greater ability to proactively address health-related issues through pattern recognition. In addition, applications can utilize other variables to add context to raw data. For example, your watch could advise you to call a doctor based on elevated skin temperature, following an extended workout and a text message that said “Worked out so hard, I’m feeling dizzy.” By proactively alerting you that you may be suffering from heat stroke, wearables can be a critical piece of preventative health care.

Appliances in the Home

The Nest thermostat, acquired by Google in January 2014, has become one of the most popular home automation appliances in the consumer market, putting IoT on the map for mainstream society. Rather than struggling to program the thermostat to change its temperature for various times of day, homeowners need only manually change the temperature. In time, the thermostat learns the homeowner’s routines and adjusts automatically. Through the mobile app, homeowners can check the thermostat’s settings and adjust them from anywhere. Other examples include fire alarms that text their homeowner if they go off when they are not home, or alerts that tell homeowners when they forgot to turn their oven off. What once seemed like a dream come true for forgetful consumers is now a common occurrence in many households. According to Gartner, the falling cost of adding sensing and communications to consumer products will mean that a typical family home, in a mature affluent market, could contain several hundred smart objects by 2022.

Crowdsourcing Data

The concept behind the “wisdom of crowds” is that outliers are eliminated or marginalized. While we typically think of crowdsourcing as a group of humans providing responses, the Internet of Things enables applications to poll a crowd of sensors or devices. Google Maps’ traffic functionality is a great example of crowd-sourced device data at work. Rather than rely on a few devices, cameras or road-mounted speed sensors, Google Maps monitors the location of drivers’ phones to create a real-time traffic map. This type of device grid can be tapped in a variety of ways to build stronger data sets, and hopefully, smarter results.

Cyberthreat Security

The Internet of Things is not solely the domain of newly connected devices; computer hardware can be harnessed en masse like never before. The “bad guys” have lead the way, with automated cyber attacks through botnets that are used to overwhelm security systems through a distributed denial of service attack, or by performing malicious actions such as click fraud on PPC ad networks or spam email delivery. Now, the “good guys” can harness distributed devices to recognize the patterns of malicious activities as they happen (instead of after the damage is already done). Cyberthreat security systems can also tap into historical data, while monitoring real-time activity, to look for older activities that may have been the first step in a multi-stage attack that can then be thwarted more easily.

If 30 Billion Devices Will Be Talking, How Will We Listen?

From a data storage and management perspective, the good news is that all 30 billion devices will not be talking to the same systems and applications. Even for one application, the magnitude of the challenge is still daunting when one considers the combination of requirements needed to store, process and manage new sensor and device data in real-time.  

Real-time analysis of real-time data can be the Holy Grail – especially when it can be run against large quantities of historical data. It’s this mixture of data volume and velocity that can be the difference between an application being merely interesting or being vital to consumers or businesses, making the choice of platform critical.  

<< back Page 2 of 3 next >>

Related Articles

The Internet of Things Heralds a New Kind of Spatial Database

Posted April 08, 2015

The Five ‘Things' of the Internet of Things

Posted April 08, 2015