Finding the right flights for travellers fast with Google Cloud Platform
By optimizing experiences on its 35 websites, Amsterdam-based Travix has been able to secure double-digit growth for the past three years. Every customer query Travix receives is transmitted to an average of four queries to Global Distribution Systems, or GDSs, each of which contacts hundreds of airlines to produce information that Travix filters and sorts in real time. The GDSs often use slow, decades-old technology, so Travix stays competitive by ensuring its own processing is done with maximum efficiency.
“Our customers expect a modern, fast system with a lot of convenience, but we often have to deal with 30 year old systems from our content providers. Our challenge is to make something that is slow and old fashioned look fast and modern. And because we’re in a highly competitive market, margins are thin, so small differences really matter. Google Compute Engine (powered by Kubernetes) gives us the flexibility to respond to demand.” — Stephan Vinkenborg, Head of Software Engineering, Travix International
Finding flights for customers – fast
Online travel ticket retail is a highly competitive industry, where speed and responsiveness to customer search queries can make the difference between success or failure in every sale. Travix depends on aging GDSs for content that can create waiting times of up to eight seconds in retrieving results, and caching alone is an insufficient solution as the content changes multiple times a day. Travix also looks to increase the speed of the customer experience by saving time in every other element of their data processing, such as content replacement, filtering and sorting.
To do that, Travix switched to Google Cloud Platform. Using Google Container Engine, powered by Kubernetes, the organization can use a flexible microservices architecture to easily manage the content changes that need to be made hundreds of times a day, each of which results in a new website being built and deployed. GCP allows the Travix architecture to scale with consumer demand and process information at servers closer to their customers, avoiding unnecessary long-distance information transfers. With a sophisticated continuous deployment pipeline that Travix has created in the cloud, they can roll out changes and developments at high speed. And because GCP is an open platform, Travix was able to run .NET Core in Kubernetes to achieve significant cost savings.
“Using Google Container Engine, powered by Kubernetes, made updates simple and very fast. Boot up time is really a factor, too, because it affects how quickly we can put together a test environment or build a new machine. Some platforms we tested took 15 minutes to do that. Container Engine takes a fraction of the time.” — Stephan Vinkenborg, Head of Software Engineering, Travix International
Enhancing stability and data analysis to benefit customers
Travix has experienced zero major issues with Google Cloud Platform in over a year of use. Decommissioning on-site servers has made their architecture much easier to maintain too. Now the company is looking to scale up business intelligence with Google Cloud Datastore and BigQuery to develop machine learning and more complex analysis to improve pricing. Major savings are anticipated for next year, as the company investigates even finer nuances of scaling capacity to meet customer demand.
“Our error rates are down and stability is up because we don’t have any unique servers any more. No stickies on servers – all that stuff is gone. With Google Cloud Platform everything is much more reproducible and predictable.” — Stephan Vinkenborg, Head of Software Engineering, Travix International