If the hardest part of the "Internet of Things" is getting to the Things, Cisco Systems is offering a lifeline.
The so-called IoT encompasses a range of Internet-capable devices that could be almost limitless: Thermometers, electric meters, brake assemblies, blood pressure gauges and almost anything else that can be monitored or measured. The one thing they have in common is that they're spread out around the world.
From a network builder's perspective, the biggest challenge this poses is backhaul, or the links between devices in the field and data centers that can analyze and respond to the data they spit out. Typically, IoT devices talk to a small router nearby, but that router may have a tenuous and intermittent connection to the Internet.
There can be huge amounts of data coming out of these devices. For example, a jet engine may produce 10TB of data about its performance and condition in just 30 minutes, according to Cisco. It's often a waste of time and bandwidth to ship all the data from IoT devices into a cloud and then transmit the cloud's responses back out to the edge, said Guido Jouret, vice president and general manager of Cisco's Internet of Things Business Unit. Instead, some of the cloud's work should take place in the routers themselves, specifically industrial-strength Cisco routers built to work in the field, he said.
"This is all about location," Jouret said. Using local instead of cloud computing has implications for performance, security and new ways of taking advantage of IoT, he said.
To equip its routers to do that computing, Cisco plans to combine Linux with its IOS (Internetworking Operating System) to create a distributed computing infrastructure for what the company calls "fog computing." It plans ultimately to build computing capability into Cisco IoT routers, switches and IP (Internet Protocol) video cameras.
Cisco announced the architecture, called IOx, at the utility-industry trade show Distributech in San Antonio, Texas. IOx will start to come out for Cisco's hardened IoT routers in the first half of this year.
To start with, the new architecture will make it easier for users to connect specialized, industry-specific systems at the edge of the network with Cisco routers, Jouret said. Different industries use many different types of connections for IoT devices, such as serial, Bluetooth, ZigBee, and Z-Wave. In the past, it's been up to Cisco to modify its routers to work with whatever interface an industry may need, Jouret said. Adding Linux to its routers changes that equation.
"Now, instead of taking many months for Cisco to do the work of integrating this interface into our router, you can do it yourself," Jouret said.
In addition, users will be able to port their current Linux applications to run on the Cisco infrastructure. They will also be able to create new sensing and control functions by writing applications using Cisco's IOx SDK (software development kit), he said.
IOx won't turn routers into full data-center servers that crunch big data. The routers, which are much smaller and lower powered than Cisco's classic enterprise and carrier gear, will carry out simpler tasks that need to be turned around quickly.
For example, if the parts of a rail car are instrumented to continually report whether they're in good working condition, a router located on the rail car could collect and process that data by itself. It would do nothing until it received a signal that showed one of the parts might be headed for failure. Then the router could report back to the cloud over a 3G or satellite link. No wide-area bandwidth would be used to send the millions of "I'm OK" messages.
On top of reducing the data burden on networks, the distributed computing infrastructure will help IoT devices operate when network connections are lost and keep enterprises from having to transport sensitive data beyond the site where it's produced, according to Cisco.
Distributing the handling of IoT data should speed up an enterprise's data analysis and decision-making, according to analyst Steve Hilton of IoT consulting firm Machnation. The security implications are important, too, he said.
"This is particularly useful in situations where sensor data cannot be transported across country boundaries for legal or regulatory reasons -- a very common issue in cloud IT deployments," Machnation's Hilton said in a blog post on Wednesday.