Challenges of IoT include big data, data analysis for enterprise

Implementing big data and IoT is difficult for enterprise IT teams due to major challenges on the network. Here's how IT can understand the relationship and prepare for the change.

Big data and the internet of things are outputs of business demands that drive changes in applications. These changes create unique challenges for the network. They also require networking professionals to stretch their view beyond the network. The challenges of internet of things and big data mean that IT needs to be prepared before jumping in to make a major change. In part one of a two part series, we'll explore the business and application drivers of big data and internet of things and share an approach to architecture to help deal with new requirements.

The IoT-big data relationship

It's an oversimplification to say that business demands are driving big data and internet of things (IoT). It's a bit of a chicken-and-egg situation. The technologies that enable big data and IoT enable new business opportunities. IoT allows new measurement and control capabilities as well as new distributed applications. For big data, it's the ability to aggregate massive amounts of data via technologies, such as Hadoop and distributed storage systems. The enormous quantities of data generated by the growth of IoT feeds the need for big data capabilities.

The internet of things is big. How big? A 2016 Ericsson report predicted that IoT will result in 16 billion devices being connected to the internet by 2021. By 2018, IoT will surpass mobile as the largest category of connected devices. With so many connected devices, it's clear that IoT will generate an incredible amount of data.

Beyond the numbers lie important questions about how the applications and data are used within organizations that access this information. A simple problem is the entry point into the network. There are plenty of IoT devices inside a data center, from IP keyboard, video and mouse devices to IP power distribution units and environment detectors. These devices normally upload data to a collector inside the data center -- or at least inside the internal network. For consumer IoT, this isn't the case. It's impractical to think that one million distributed consumer IoT devices will enter the network from a single entry point.

There's not a one-size-fits-all solution to the problem of data collection. A significant portion of the design is dependent on where the data analysis will take place. In most big data examples, the information is centrally located and processed as part of a centralized application.

Data analysis creates other challenges of IoT

In enterprise IoT, a single point of entry is feasible. The devices are in contained and trusted environments. Industrial IoT, however, may prove a challenge. IoT devices can potentially disrupt operations based on invalid readings, such as an incorrect temperature. Plus, the sheer volume of wearable data makes industrial and corporate IT systems inefficient.

Data analysis is a larger-scale challenge, too. For example, the goal of a wearable device could be to alert medical staff to anomalies in vital signs. In that case, waiting for a centralized batch process to run across a large data center to find irregularities is extremely inefficient. An alternative design is to place anomaly detection closer to the network's edge.

Cisco's big data architecture includes the concept of a fog node, which sits on the edge of a network or cloud provider's point of presence. The term fog is a hat tip to the concept of the cloud. The node connects to a network entry point either at the customer's site or within Cisco's Jasper cloud network. The proximity of the fog node to the data allows fast data analysis. The node detects anomalies and acts upon them as information is collected, and optimizes the data set before it uploads to a central database.

Ericsson's 2021 prediction includes 1.5 billion IoT devices with cellular connectivity. It's safe to say that mobile carriers will look to capture some of the revenue associated with data collection from these devices. Cisco-owned Jasper provides overlay services for collecting data from devices, including from autonomous vehicles. Expect third-party products and provider-hosted offerings to spring up around this market.

Security will be a factor, too, though the extent of that concern will vary. In use cases where IoT devices hold personally identifiable information or that produce video content, device security will be a priority.

Security is another advantage to this distributed data-collection-and-analysis design because edge data collection devices act as security endpoints. The endpoints can detect security events and kick off a workflow for either isolating wearables or taking corrective action, such as flashing the device's firmware. The edge nodes allow scaling device configuration validation.

These challenges of IoT and big data create problems for enterprise IT teams. See how the nature of the data and distributed databases can help close the knowledge gap in part two.

Next Steps

Discover the changes that IoT can bring to your network

Avoid network stress with IoT edge analytics

Address IoT security concerns

Dig Deeper on SDN and other network strategies