analyticsweek_black-300x59

One of the most significant points of evolution in the modern data landscape is the inexorable transition of a centralized model of data management to a decentralized, distributed one. The harbingers of this change include:

  • Big Data: The sheer amounts of data, and speeds at which they are generated today, result in frequently occurring situations in which data is engendered outside of the enterprise, as opposed to within it.
  • Cloud Ubiquity: The copious quantities of data have made the cloud an increasingly viable medium for accessing and processing them, particularly in terms of big data, social media and sentiment analysis, and the integration of external unstructured data with internal, structured data.
  • The Internet of Things: The merging of big data and the cloud have resulted in the transformative power of the IoT, which emphasizes the increased appositeness of mobile technologies in accounting for this convergence.
  • Sensor Data: The continuity of data created by remote sensors and the analytics rigors associated with gleaning insight from such data typify the combination of the foregoing factors.

According to Ryft Vice President of Engineering Pat McGarry, the emergence of these developments has created a common problem for organizations today: “There’s just so much data growing exponentially, and if there’s one problem we have it’s how the heck do we analyze it to get the answers we need quickly enough?”

The solution is partly based on either supplementing or replacing centralized methods of transforming, transmitting, and analyzing data (which has considerable infrastructure, network, and temporal costs) with edge computing, in which analytics are performed at the cloud’s edge closer and quicker to the devices that need them.

It also involves extending the utility of the cloud’s edge via the phenomenon known as cyberforaging–the sharing of computational resources between mobile devices to bolster the advantages of edge computing and account for the challenges of analyzing sensor data.

Sensor Data Analytics
Currently, the business value generated by cyberforaging is most notably found in marketing. But as McGarry pointed out, the greater significance of this amalgamation of big data, cloud computing, edge computing, and mobile technologies is that “More importantly, it’s a way to analyze sensor data in real time. That’s really what’s happening there.” The tailored marketing efforts of real-time sensor data analysis arguably exceed any others, since they are predicated on actual behaviors of customers as they are engaging in them. Contextually-based advertising systems can leverage the sensor capabilities of mobile devices and cloudlets by determining what marketing materials are sent, in what order, and in what context. Typically this process begins with designing enterprise apps that customers download to their phones. As McGarry noted, when “you walk into a store, you have their app and you’re on Wi-Fi, you’re connected right there at their store. They can literally talk to you directly then, and they can offload things about you, directly analyze them, and send the results back.”

Power of the End Device
In some respects, cyberforaging is possible due to the increased power of end devices. Whether such devices are mobile (in the form of tablets, smart phones, or laptops) or otherwise, the increased computational power found at the edge of networks makes performing computations and issuing the results there at the cloud’s fringe advantageous. Cyberforaging is largely predicated on the sensor capabilities of mobile devices, which can be used to detect additional devices nearby (known as cloudlets) to offload computations or even vital data preparation for performing analytics. Cloudlets are any variety of devices with servers and internet connections that function between end points and the cloud for the purposes of offloading. They also work without an existing internet connection—once their resources have been provisioned. The benefit of cloudlets is that they are closer to the fringe of the network and enable preparation and analytics without having to constantly transmit data generated at the cloud’s edge to a few centralized locations.

Computational and Preparation Off-Loading
The power of mobile devices (such as smartphones) is greatly augmented by cloudlet-based cyberforaging. Heavy analytics jobs can be performed on the fly with reduced costs and infrastructure compared to those for maintaining centralized data centers. Remote data preparation can be used to facilitate ETL for BI and analytics purposes, or in some instances, be avoided altogether. These cyberforaging benefits also pertain to greater flexibility and autonomy for data access while furthering the self-service movement. The classic example of cyberforaging is found in telecommunication style applications, but the potential for this phenomenon to revamp sensor data analytics and analytics at the edge of networks on the whole is substantial. “It’s being done in some places, but it’s not being done to the extent it needs to be done,” McGarry remarked. “It has to be this way. There’s so much data being generated now at the edge of these networks; you have to do something. You can’t ship it all to a central location anymore. You’ve got to ship it somewhere else, or you’ve got to process it right there as locally as possible.”

Real-Time Mobile Analytics
The concept of cyberforaging can be as simple as utilizing a popular search engine’s cluster to perform an operation and send the results back to a mobile device. It’s centrality to the way it solidifies the distributed paradigm of data management lies in the way it reduces physical infrastructure, network and bandwidth strains, and costs associated for analytics. End users can reap the benefits of distributed analytics without having to wait for centralized approaches and the bottlenecks they create. Additionally, cyberforaging increases the power of mobile devices (or any device at the edge of the cloud), their speed, and the autonomy that is part of the self-service movement in data management. These advantages will resonate even further as the IoT continues to gain credence. Moreover, they help to bolster the utility of sensor data analytics and make them less complicated and more viable for organizations seeking to exploit this technology.

“Throughout history, knowledge has been power,” McGarry stated. “We’re generating so much data, and the trick is getting knowledge out of that data to use it as a competitive advantage over somebody else. That is really at the crux of it, and all of this stuff is related directly to that. So I think businesses in the end have to figure this stuff out.”

READ MORE