Fog Computing Feature
November 12, 2014

IoT Data Needs Priority Treatment

The world around us is filled with data. In no uncertain terms, it's plain old everywhere. From the data that powers voice over Internet protocol (VoIP) connections to the data that lets us send pictures of cats around the Web, it's hard to get away from data. But while the rapid growth of data is already putting a strain on systems, one point of technological growth—the Internet of Things (IoT)--will generate much more data than most saw coming. It's this data that should be looked at more closely, some assert, and even given priority treatment in the long term.

While the IoT doesn't engage in passing around photos of cats and assorted whatnot, it rather passes around climate data, temperature, humidity, location...a host of informative bits about the surroundings and positions of the sensors in question. With the unique nature of the data the devices comprising the IoT gather comes a host of equally unique considerations in terms o f memory and data storage requirements.

Since memory can be found in a host of devices—from sensors to smartphones and beyond—the term is a bit broad, according to IHS analyst Cliff Leimbach. But most parts of the IoT usually have to send data back to a central hub, or otherwise do something with the data gathered, and that makes the prospect a little murkier. While Leimbach points out that memory requirements will be much lower if the device in question proves little more than a relay back to a central hub, it's more complex uses that may shake up the equation further.

The market for dynamic random-access memory (DRAM)--commonly seen in such devices—has been on an upswing for the last few years, growing at a rate of about 30 percent a year. DRAM is also not losing much capacity to the tablet market; reports suggest that only six percent of DRAM bits are going to tablets, meaning there are other uses afoot as well. Leimbach notes here that, should a device really want to “make a dent,” it would need to have plenty of power behind it and that will likely call for DRAM. NAND Flash is likely to also be called upon in this field, particularly should the device be doing much more than temporarily holding data to route elsewhere.

But given that Flash is much more expensive than regular hard drives, some companies are already starting to ask—and here, Enterprise Strategy Group analyst Mark Peters seems to believe that more of these decisions will have to be made—if keeping Flash on hand is worthwhile. Here, Peters notes that it “...really depends on what you are going to do with all of this data that IoT generates.” For instance, Peters notes that monitoring emissions from passing traffic at a toll booth really doesn't require speed, but trying to catch a stolen car within a few hundred yards of the toll booth would.

The key, at the end of it all, is the application. It's not the data, but rather, what's done with it, and that means making IoT data a priority is going to be important. Without a proper handle on the functions desired, there will be either overpurchasing of memory, and therefore waste, or underpurchasing and lost opportunity from inferior systems. What's done with the data determines what's needed to handle it, and knowing how to handle it will help ensure the best system for the job is on hand.

Edited by Maurice Nagle

Back to Fog Computing
Comments powered by Disqus