Artificial Intelligence: Splunk at Cox Automotive

splunk_glasstable_process

Splunk ITSI

At its annual user conference, .conf, Splunk announced new versions of Splunk Enterprise, Splunk IT Service Intelligence (ITSI), Splunk Enterprise Security (ES) and Splunk User Behavior Analytics (UBA).  These products leverage machine learning to speed-up and facilitate extracting insights from machine-generated data.

Splunk was founded in 2003 and brought “big data” to Wall Street’s attention with its 2012 IPO. It has always focused on machine-generated data and its platform captures, indexes and analyzes real-time data in a searchable repository, on premises or in the cloud. Machine learning extends the Splunk platform by adding outlier and anomaly detection, adaptive thresholding and predictive analytics capabilities, applying over 25 commonly-used machine learning algorithms or custom algorithms to build data models that forecast future events.

One of the more than 12,000 customers using Splunk to extract value from the avalanche of machine-generated data is Steven Hatch, manager of enterprise logging at Cox Automotive, a subsidiary of Atlanta-based Cox Enterprises. With multiple brands such as Manheim, Autotrader and Kelley Blue Book, Cox Automotive is changing the car buying and selling business, helping people buy and sell cars from their homes, offices and mobile devices.

“Cox Automotive represents the overall lifecycle of a car,” says Hatch. “Whether it’s a dealer or a consumer there are specific functions within that lifecycle that generate a lot of data, up to 2 terabytes a day. Machine learning can be applied to this data to understand better how the marketing campaign is impacting web traffic or how dealers leverage parts and services. We have all of this digital exhaust over the lifecycle of a car and machine learning can draw a picture and analyze events, trends and activities we didn’t even have thought of before. That’s the beauty of machine learning.”

Manheim, for example, operates more than 75 car auctions across the U.S., selling thousands of cars daily. These auctions are conducted via a bidding process that can be done locally or online via simulcast real-time auction video streams, using microphones and high-definition cameras installed in each auction lane. Massive amounts of data are generated by an audio-video scanner tool that pings and polls devices at each auction lane every 30 to 45 seconds. Cox Automotive must monitor all network traffic across many auctions because failure of any device on any lane affects their customers’ experience and the company’s bottom line.

In the past, Cox Automotive encountered uptime and application stability challenges during its auto auctions, but had little visibility into the root cause. When a problem occurred, its operations teams lacked visibility into whether the disruption was broad across the network or isolated to a single lane that needed investigation. It also lacked the ability to prioritize incident investigations and needed real-time insights into the performance and availability of each auction lane.

Implementing Splunk IT Service Intelligence (ITSI) helped drive down key metrics such as mean-time-to-identify the root cause of an incident and mean-time-to-resolution. Now, if an incident with a camera, microphone or other device occurs, staff members get an alert within seconds, can troubleshoot quickly, identify the issue rapidly and determine the exact location for an auction technician to minimize disruption. Moreover, using advanced analytics and machine learning, the IT staff can predict outages and can even monitor equipment degradation for proactive replacement. “Splunk ITSI brings visibility into our simulcast application that we never had before,” says Hatch.

The benefits of applying machine learning tools to machine-generated data, however, go beyond improved IT operations. Now that most business activities are online, there is a lot of business-related data that could be analyzed, not just the data that the IT infrastructure generates. Business users at Cox Automotive, says Hatch, “can improve their business decisions because they now have insights they never imagined before, or have data to support going forward with a specific initiative.”

The fundamental benefit for the business is in the act of the aggregation of all the massive amounts of data machines generate, something that was not possible before the advent of cloud computing. “Six years ago this would have been a monumental effort to centralize all of this data because we didn’t have the luxury of having compute and storage in the cloud the way we have it now,” says Hatch. Now, all of this valuable data can be accessed and analyzed by the people of Cox Automotive who could use it to make better decisions. Concludes Hatch: “Big data is what it’s all about. Big data is not the data that is rigged to the hard drive. Big data becomes big when it can be shared across the business to make the right decisions.”

Originally published on Forbes.com