If you thought the in-house data silos of the client-server era were a nightmare, get ready for the hairball that the internet of things could engender. As tens of billions of “things” — sensors, machines, mobile devices — get connected to the internet and to each other, there will be huge value in the data they generate across applications ranging from agriculture, to senior care, to public works.
But there’s a lot of heavy lifting to do first. We know we can collect the data — but how to process it, analyze it and share it securely in a way that makes sense and doesn’t give people the heebie-jeebies? These are topics we’ll dive into at Gigaom’s Structure show in June.
Some parts of the internet infrastructure are ready for this IoT onslaught already. “If just 15 to 18 percent of all dark fiber is lit up now, there’s still tremendous amount of capacity available,” said Chad Jones, VP of LogMeIn’s Xively IoT business. “It’s not that we need the pipes, but we do need a ton of infrastructure atop the pipes. The next thing is we’ll need a bunch of clouds meeting different requirements from all these data sources and we’ll need business intelligence and stream processors etc.,” he noted.
And there will be a need for myriad data aggregation points — they could be small 10-rack data centers or smart routers situated close to data sources. Mark Thiele, executive vice president of data center technology at SUPERNAP, (pictured below) who will also speak at Structure, said there’s going to be a whole lot of tiering going on which distributes smaller collection points that communicate with massive core data centers as needed.
Overcoming structural barriers
And, if the value in big data lies in bigness, people will need to overcome their need to horde their data and instead share it — in blinded or anonymized form — to get the biggest bang. Demographic or health data from one block in the neighborhood is of limited use. To truly fuel demographic research, that data must be combined with city-wide, state-wide and nationwide data.
And the main barrier there is psychological or sociological, not technical. “If you think how hard it is for departments of one company to collaborate and share information, and then look at how hard it is to get agencies within one government to do the same, and then getting governments of different countries to cooperate, you see the problem,” Thiele said.
To foster interoperability and sharing there also has to be a way for companies that collect their own data to monetize it. Forward-thinking companies need to come up with a standard unit of data to foster the ability of that data to be traded, Thiele said.
6Fusion already does this in the Infrastructure as a Service market where it lets customers compare like units of workloads across deployment models. It does this now with what it calls a Workload Allocation Cube (WAC) that factors in compute, storage, disk, WAN I/O and LAN I/O in a single unit of measurement.
The same could be done with data — cutting it up into widely-recognized measurable units that can be bought and sold like any other commodity, said Rob Bissett, senior vice president of product management for 6Fusion.
Bissett thinks tradeable data units will be a common topic of conversation in 3 to 5 years after some of the infrastructure stuff gets worked out, but it will have to happen. Let’s face it: generating a lot of data that lies fallow in various cloud islands is of limited use. The real value will lie in the ability to parse and manipulate it so that roads and bridges get fixed before they fall apart and jet engines can be attended to before real problems arise.