At Priceline, part of Booking Holdings, Inc. [NASDAQ: BKNG], they believe every trip is a big deal to their customers. Whether it’s a much-needed getaway, wedding, reunion, graduation, or rooting on a favorite team—those are the moments that nobody should miss out on. Their mission? Be the best travel dealmakers in the world. They provide their customers with smart and easy ways to access the very best deals available on hotel rooms, flights, rental cars, vacation packages and cruises, so they can be there for the moments that matter the most to them.
Cloud Migration and Deployment Pattern Changes
Priceline originally built a custom deployment solution, which they successfully used to deploy services to their on-prem environments. This worked because the on-prem environments were relatively homogeneous and Priceline kept to a strict set of deployment patterns.
When the team at Priceline decided to move applications from on-prem to GKE and simultaneously wanted to develop new deployment patterns, they knew they had to make a change to their tooling. Priceline’s DevOps team quickly realized these initiatives were out of scope for the homegrown tool and began the search for a new solution.
“The work to create new deployment patterns was not something we were going to realistically accomplish on our own.”Jason Solis, CD and Automation Architect
Using their existing solution, the team would need at least 3 engineers for 6 months just to implement canary deployments in GKE.
“It would have been a train wreck if we had tried to move to GKE by ourselves. We didn’t have 10 years to stand everything up, we needed to move fast.”Jason Solis, CD and Automation Architect
Technically, the team could have repurposed their homegrown pipelines for the cloud, but Priceline wasn’t willing to waste the time and effort developing and maintaining their own software when they could find a tool that specializes in what they needed.
“Homegrown is fun for smaller-sized companies, but doesn’t work for enterprises.”Jason Solis, CD and Automation Architect
Modernization With Little Effort
Priceline chose Harness to modernize its technology stack. With Harness, Priceline was able to adopt advanced deployment patterns with little effort. The canary deployments that would have taken Priceline 6 months and 3 engineers to set up took 1 developer a couple of days using Harness.
“Any custom tool we would have built would have been extremely limited, and wouldn’t have scaled.”Michael Bazos, Software Engineer
Harness also made the transition to GKE easy by providing things like secrets management, and by allowing developers to easily create new pipelines for any new Helm chart.
So far, the move to GCP and Harness has worked out great. Harness helped Priceline quickly and easily build deployment pipelines from scratch for its migration to GCP and GKE. Currently, 45% of Priceline’s applications are hosted on GKE, with the goal being to migrate 100% of the applications.