Kafka in the Cloud: Why it’s 10x better with Confluent | Find out more
Sarvant Singh, vice president of Data and Emerging Solutions at Penske Transportation Solutions, will tell you that Penske is more than just the yellow trucks you see on the roads. And rightly so.
“We deliver a full range of innovative transportation and logistics solutions that are vital to the success of the company and the customers we serve,” Singh said. “We’re an information-intense business, where a lot of information is getting exchanged between our customers, associates, and partners.”
This also means it has become more important than ever for Penske to collect all that data in real time to make informed decisions based on that data—and drive better customer experiences and business efficiencies.
“In our business, vehicle uptime and supply chain visibility are critical,” Singh said. “The real value is in ensuring that whatever our vehicles are transporting reaches its destination on time. Uptime is the number one thing which our customers look towards us for. To drive that, we have multiple digital, innovative solutions already in place or are in the process of implementing.”
With technologies built around data streaming, this year alone, Penske estimates to be successfully preventing over 90,000 vehicles from encountering roadside breakdowns.
Each Penske vehicle has hundreds of sensors supported by thousands of sub-sensors, and all that data is being collected in real time. Being able to understand the real time sensor data is imperative for providing a better experience to their customers, drivers, fleet managers, and even technicians, according to Singh.
Confluent is one of the primary mechanisms to process real-time event data at Penske.
“We process around 190 million IoT messages a day using Confluent as our messaging queue. Then we run those messages through our proactive diagnostics engine—an AI engine which helps us predict when a vehicle may fail,” Singh said. “With that information, plus the knowledge of where the vehicle is right now, we can call the customer and ask them to bring the vehicle in, or schedule the maintenance that’s needed.”
Today, Penske uses telematics data (including location data and engine fault codes) to keep vehicles healthy and streamline functions, including:
Expedite roadside assistance: Telematics data enables swift issue diagnosis and service dispatch
Preventive alerts: Telematics data triggers alerts for proactive maintenance scheduling
Easy fuel tax compliance: International Fuel Tax Agreement (IFTA) integration automates data transmission from telematic devices, eliminating paperwork or administrative nightmares.
Efficient repair process: Penske’s electronic driver vehicle inspection report (DVIR) integration streamlines defect reporting and resolution reducing paperwork and saving time
To learn more about how data streaming with Confluent can power your AI initiatives, start exploring our AI-specific use cases resource hub today.
Once Penske made the data streaming decision, they knew they wanted to opt for managed services for Kafka—because they didn’t want to deal with Kafka management woes.
After working with a Confluent competitor for a while, Penske decided to move to Confluent.
“Today, the way we respond to our customer needs has completely transformed because of the availability of real-time data in Penske’s data platform, and Confluent plays a big role in that,” Singh said.
Here’s a look at some of the key reasons behind Penske choosing Confluent:
Managing Kafka is painful and requires significant engineering prowess. Penkse would have needed a significant number of FTEs to support open source Kafka. As a fully managed service offering, Confluent handles all the complexities around Kafka
Your data streaming platform should scale up and down rapidly and easily to meet customer demand, but open-source Kafka is not easy to scale. With Confluent Cloud, scaling is much faster than with Apache Kafka—and doesn’t involve the capacity planning, data rebalancing, or any other typical operational burden that goes into scaling data infrastructure.
Confluent goes above and beyond open-source Kafka by providing a set of enterprise-grade tools and capabilities (including Connectors and Stream Governance) for a complete data streaming platform.
“I’m a big advocate of open source software, but we're not going to put mission-critical applications on an open source tech,” he said. “Enterprise-grade applications require enterprise level support—and Confluent's business value has been clear as we scale our IoT data. We see reduced downtime, improved time to market, and we don't have to provision anything. Scaling happens at the speed of thought—as soon as we're ready, we know Confluent is.”
Truck fleets deliver approximately more than 70% of the freight in the U.S. and the supply chains are extremely lean. Customers operating these vehicles need to know where, when and how their vehicles are operating in real time.
Similarly, those customers who are waiting for their freight to arrive also need to know where and when that freight will arrive or if it will be delayed. Disruption to the movement of freight by truck, air, rail or ship, has a domino effect and can cause production delays at a manufacturing plant or stock-outs in a local grocery store as we all saw in the pandemic.
Real-time data helps trucking companies and those who depend upon them in the supply chain make adjustments in real time.
That’s why driving use cases like predictive maintenance or vehicle remote health monitoring are mission critical for Penske. And Confluent supports IoT-related use cases like real-time vehicle location tracking and proactive diagnostics to name a few, Singh said.
“Our fleet consists of more than 400,000 vehicles, but those vehicles aren’t at our locations. They’re on the road. That’s where problems can happen. Before we used streaming technology, if a driver was experiencing an issue, finding out where that vehicle was located and what was happening to it was challenging,” he said.
Understanding all the sensor data from vehicles is incredibly valuable to Penske. It allows Penske to provide a better driver experience by ensuring each vehicle is in the best working condition and also makes their technicians’ jobs easier by proactively evaluating potential issues—which is especially helpful with the technician shortage the industry is facing.
“Being able to understand sensor data in real time leads to significant business impact,” he said.
Penske is in the process of onboarding the rest of its fleet to use Confluent— currently, there are about 165,000 vehicles sending data to the platform.
“As more electric vehicles join the fleet, we’ll see a lot more data generated at the vehicle. We’ll have electric chargers sending us data. Overall, there will be a multi-fold growth of data streaming at Penske,” Singh said.
Penske's journey with AI dates back to 2015, when the company started looking at ways to automate routine and low complexity tasks, while increasing the availability of their services.
Penske implemented a virtual assistant called Erica to help existing customers with confirming or modifying their rental reservations, Singh said. For an existing customer, this virtual assistant identifies them by their phone number, and can quickly check their reservation information and make changes if needed.
The result? Improvement of the overall performance of the Penske call center.
“With a fleet of more than 400,000 vehicles, 40,000 associates, and thousands and thousands of customers, there's a lot of information exchange happening between all these entities. Our goal is to make every experience a better experience,” he said. “We are looking at how we can totally transform a process using something like a conversational AI—and call centers are a very logical area for us to start.”
Today, Penske leverages Confluent to power use cases like real-time vehicle location tracking and proactive diagnostics to name a few, Singh said. Confluent helps with bringing the right data to the right place at the right time—by routing relevant data streams anywhere they're needed in the business in real time for reliable, trustworthy use.
“We’ve very successfully implemented AI to reduce the repair time for our vehicles—and in some cases, we can predict a maintenance event before there is a need, which has increased uptime for our customers,” Singh said.
But AI implementation comes with its challenges.
“There’s a reason why it is still called an emerging technology. The future looks very promising, but there’s more to be seen. I think the technology needs to evolve to provide a more human-like experience to the users of the technology,” he said.
And now there’s ChatGPT, which companies are exploring due to the immense potential it has to revolutionize business automation. But like any other technology you have to be cognizant of risks—including data accuracy (hallucination is a big challenge) and cybersecurity (handling of PII data), Singh said.
“In a platform which is offering ChatGPT-like capabilities, traceability, explainability, anomaly detection, and configurability, are very important,” Singh said. “ Configuration for risk tolerance is going to be the key. That’s because risk tolerance is not consistent across various business processes and functions. Without that, no implementation in an enterprise setting can move forward.”
And Penske is constantly on the lookout for expanding AI technology usage to drive decisions within the business, he said.
“Our business is built on three core values: commitment to our customers, dedication to excellence, and innovative thinking. These values guide the majority of our decisions around our initiatives,” he said.
Transportation technology is evolving fast, Singh said, and we can expect to see the industry transform as we see more digital innovations—including development of electric and autonomous vehicles, adoption of Internet of Things (IoT) and real-time tracking, and integration of data analytics in the business processes to optimize asset management, routing, inventory management, and demand forecasting.
As autonomous or semi-autonomous vehicles roll out, the customer experience, the digital experience inside the cab, is going to shift. Autonomous vehicles do a lot more data processing, and computing on the edge will keep getting bigger. Real-time monitoring of what's inside the vehicle like temperature of products inside a refrigerated unit will become more prevalent.
“The combination of AI, remote sensing technologies like Lidar, edge computing, and real-time data processing will expedite the race to the “left” and the business and technology architectures will need to evolve accordingly,” he said.
To learn more about how data streaming with Confluent can power your AI initiatives, start exploring our AI-specific use cases resource hub today.
This blog explores how cloud service providers (CSPs) and managed service providers (MSPs) increasingly recognize the advantages of leveraging Confluent to deliver fully managed Kafka services to their clients. Confluent enables these service providers to deliver higher value offerings to wider...
With Confluent sitting at the core of their data infrastructure, Atomic Tessellator provides a powerful platform for molecular research backed by computational methods, focusing on catalyst discovery. Read on to learn how data streaming plays a central role in their technology.