Best deal for Priceline

Highlights

• Aiven was key in the digital cloud-native transformation of Priceline
• Priceline’s savings come from not having to maintain the system and from Aiven’s business model where data costs are not charged.
• Priceline continues to expand and scale, and that’s easy to do with Aiven.

The travel industry has seen not just steady growth, but also enormous change and upheavals in the last few years. Upheavals are, of course, represented by COVID-19, which resulted in far reduced numbers of travelers, both in terms of business trips and holidays.

But the changes are multiple. Priceline is a good example of the rise of companies offering new pricing models and increasingly ubiquitous service access. The new pricing models are made possible by the heavy use of pricing and availability data generated by airlines, hotels and rental cars.

The same data-driven approach drives online travel agencies to the cloud in order to serve customers better and faster, explains Sachin Menon, VP Data & Marketing at Priceline. “It all comes down to how good a deal the customer gets when they do a search, and that governs how likely they are to make a booking.”

The challenge

In 2019, Priceline was already taking advantage of event-driven architecture by running their own Apache Kafka servers in 2 data centers, serving 4 regions. But the benefits of cloud-based architecture were clear even then. Instead of investing in machines, premises and electricity, going into the cloud means investing directly into software and capacity. Cloud computing brings services closer to the consumer, increasing access and speed. Disaster recovery and security are also enhanced.

It was also critical to get valuable insights from the data in real time to enable data-driven decision making and quicker time to market deployments.
Priceline could have opted for a self-managed Apache Kafka cloud solution. But although Kafka is a great open source answer to many business cases, it’s definitely not effortless.

With their on-premises solution, Priceline already found themselves spending large amounts of time maintaining their data pipelines and updating the servers instead of working on the algorithms and products that were their core expertise. When they decided to move to a cloud environment, they began to search for a partner who could provide a fully managed and hosted Apache Kafka solution and turned to Aiven.

The solution

Since Priceline’s data was locally stored, their first step would have to be migrating it into whatever cloud they chose. That would be the expensive part; most cloud providers charge for both egress and ingress data. Aiven doesn’t, though. By choosing to go with Aiven, they recognized significant cost savings right off the bat.

Projects of this magnitude are never easy, and it took Priceline about a year to fully achieve their cloud-native transformation with Aiven. One thing that helped was a factor they had already recognised during the vendor selection process. “Aiven’s Solutions Architecture group fully focused on finding the best solution for us,” says Sachin. “They weren’t looking at it from a sales perspective at all, not trying to push Aiven’s offering. Instead, they would propose a course of action, and then we’d evaluate whether Aiven could support it.”

Together the teams planned out the migration, and every now and then they had to wipe the slate clean and think again. Due to the approach they took, some features that the Priceline team really wanted to include weren’t available from Aiven, but they’ve now been incorporated into Aiven’s roadmap.

The data was migrated en masse onto Aiven’s servers and replication set up, so that instead of two data centers, they now have 4 regions with 2–4 data centers each. Without having to lay a single new brick, they’re now able to serve a much wider range of customers at higher speeds.

The outcome

Today, Aiven for Apache Kafka is a central component of Priceline’s streaming data pipeline, and Aiven is one of the key partners in their data strategy. Sachin is very happy with the situation now.

“With Aiven, we now have the flexibility to enormously scale to accept new data and new users and ingest everything into the analytics layer,” Sachin says. “We can get insights into the data in less than two minutes, which has helped us enable a more efficient reporting timeline.”
Sachin is also happy with his team’s reduced workloads. “With a healthy infrastructure and stable pipeline, and the fully managed platform that Aiven provides, there are fewer production support issues that my team has to look into. It’s just working without surprises.”

But on a higher level, Priceline’s team can now focus on their own core task: improving the product and increasing conversions, resulting in better deals for their customers. They don’t need their own in-house data infrastructure team, which frankly saves them lots of money.

Priceline is reviewing opportunities to scale up their operations with Aiven. “Our business is growing, and we’re starting to discuss additional volumes. We’re also looking into enhancing our understanding about the health of our clusters. And I’m thinking about auto-scaling, maybe adding some AI capabilities to leverage historical data.”

So what does this all mean for the end user, using the Priceline app to buy flight tickets? “Our goal has always been to provide the best deals to our customers. The more we add analysis capabilities, the better the deals we can provide. Aiven is helping us achieve that.”