Travel technology giant Amadeus' move towards open source and the cloud can't solve its biggest issue

Airplane

The travel industry is dealing with a look-to-book ratio which remains stubbornly at around 1000/1. Even a shift in technology strategy can't bring this down for Amadeus: all they can do is serve search faster.

Share

Travel technology giant Amadeus has shifted to being a cloud-native company, and is using Red Hat's OpenShift Platform-as-a-Service product to avoid cloud vendor lock-in and deploy its core applications wherever it sees fit.

The reality the travel sector is facing up to is that for every thousand online searches, on average only one booking is made.

Dietmar Fauser, Amadeus IT group's vice president of architecture, quality and governance, is realistic about what his organisation can do to face up to the look-to-book-ratio problem: "The look-to-book ratio is not coming down, so the goal is to offer cheap systems to sustain from a decent cost basis this massive search traffic. The end game is to get bookings, so the better you serve search the better bookings."

Read next: Red Hat looks to make hybrid clouds more useful with openness and flexibility

Amadeus hosts the technology which powers a huge share of worldwide travel transactions, linking airlines, hotels and travel providers with corporate and leisure clients. As Fauser put it: "Amadeus is the distribution system to link the content providers into a single system where you can visit all of the individual providers."

Red Hat OpenShift

When it came to shifting to the public cloud Amadeus didn't want to be locked into a single provider, like their main rival Sabre has. "As far as I know about Sabre they have also gone onto the cloud but they took a different approach to go with AWS," says Fauser. Instead, Amadeus will use AWS solely as an infrastructure partner to remain flexible.

By running on OpenShift Fauser says: "We took a fundamental tech decision to make our apps independent from the underlying infrastructure, to have an applications platform that could run on-site, be distributed over multiple datacenters, or be accessed through public, private, or hybrid cloud platforms."

As transaction volumes continue to increase for Amadeus the company decided that distributed computing was the only way to manage the scale.

"Metasearch engines changed the game," Fauser says. Most of these transactions originate now with the metasearch engines, giants like Kayak.com, Expedia and more recently, Google's own flight search.

The Amadeus booking platforms process nearly 500 billion travel-related data requests per day, "so it becomes more and more evident that in order to offer the best user experience with minimum latency, it is economically and technically reasonable to distribute these massive workloads as close as possible to where the transactions originate," he says.

Read next: Amadeus aims to spot system errors before they happen with Netuitive

From a technical perspective, metasearch created a huge amount of complexity. As each search incorporates a range of dates, providers and systems "executing these queries from an IT point of view becomes nasty."

Amadeus tried to piece together a home made framework, sitting in its own environment, built on Linux to deal with this increased parallelisation within the system, but quickly realised they would need something more robust when it came to deploying it in the cloud.

This is where OpenShift really shined for Fauser and his team: "It figures out how to place this workload on any given environment."

Open source

Another, softer benefit of shifting to open source technology is in attracting the best tech talent into the business.

Fauser says: "You can attract the top talent when you offer them the possibility to contribute and to discuss issues with Google and Red Hat engineers and access new features."

Read next: Airline IT giant Amadeus buys Accenture's Navitaire for $839m; reveals big data architecture

Finally, Fauser acknowledges that a managed approach to open source has a cap, and if he could manage it inhouse, he would.

"We are cost conscious and open source is not free, you have to invest in internal capabilities," he says.

"We pay money to Red Hat for enterprise support but if this amount becomes too much at one time I will think about having a team of engineers that maintains it for us, so there is a natural capping in this type of model. If I had more than 20 engineers I could cover it myself [...] there is room for a commercial model but you would not expect the aggressiveness of an Oracle here."

Find your next job with computerworld UK jobs