peshkov - Fotolia

Amadeus turns to NoSQL to answer complex travel questions

Travel software and services company Amadeus has employed a NoSQL database technology to help travellers find answers to complex travel queries

Amadeus, the travel technology company set up by a group of European airlines in 1987 to enable travel agents to carry out flight ticketing online, has turned to a NoSQL database technology to enable travellers to ask complex questions about their journey.

Olaf Schnapauff, CTO of global operations at Amadeus, gives the example of someone with a €600 budget who wants to go somewhere warm during the dark northern European winter, and wants to make an enquiry expressing those needs.

Schnapauff is a scuba diving instructor as well as a technology executive, and joined Amadeus a year ago. “I have travelled a lot, often to remote locations as a diving instructor, so I have always been fascinated by what makes it all work,” he says.

Part of what makes it all work, from his company’s perspective, is located in Amadeus’ own datacentre in Erding, but the organisation’s IT strategy is to evolve its cloud services, in part leveraging the Google Compute platform, says Schnapauff. “We run 2.3 billion transactions a day, we board 747 million passengers a year and process four million bookings a day,” he says.

“At the lower end of the stack, we run our own private cloud infrastructure as a service based on VMWare’s OpenStack distribution. We also run in production on Google and plan to run on other public cloud infrastructures. And we provide platform- and software-as-a-service layers on the upper end of the stack that can run on any public cloud.”

The company chose NoSQL database MongoDB to help it build an “instant search” application that can browse billions of travel options across multiple criteria in real time. Travel site Kayak is using Amadeus Instant Search technology to increase conversion rates from “looking” to “booking”.

“Online channels have transformed the way people plan and shop for trips,” says Wolfgang Krips, executive vice-president of global operations at Amadeus. “They want to be inspired by travel choices and they want to explore, compare and buy now. Providing instant results to complex search queries is daunting and requires cutting-edge technology.”

Swiss army knife of NoSQL

Amadeus is using MongoDB as the “Swiss army knife of NoSQL technology”, says Schnapauff. “It is a good fit for when many documents have to be processed to perform a query that is complex. We also use Couchbase as a key value store and as a caching layer for large query volumes when the search is across only a few dimensions.”

Amadeus also uses relational database technology in the form of Oracle and MariaDB for more traditional bookings.

“Initially, Instant Search was developed on an internal NoSQL database, but it could not handle the increase in load, while relational databases were not designed to cope with the scale and agility we require,” he says.

“Instant Search teams found other teams at Amadeus were benefiting from the use of Mongo, which is a gold standard technology in the company’s technology forum, adds Schnapauff. “We were able to scale Mongo by distributing the data across multiple shards using the Wired Tiger storage engine. The security and compliance features provided by MongoDB are critical for us.”

Read more about Amadeus

  • Amadeus is adding to the business intelligence capability it has been building since 2013. Its head of travel intelligence says the industry is at the start of a data analytics revival.
  • Wolfgang Krips, head of global operations at Amadeus, on how automation and a skilled team work in unison to provide high-availability IT services.
  • Amadeus operates a major datacentre in Erding, Munich, which provides ticket distribution and IT services to the global travel industry. Read the interview with Gerry Smith, vice-president of operations.

As CTO, Schnapauff’s strategic priority over the next few years is “to complete the journey to hybrid cloud”.

“We have applications in production on Amadeus cloud services with containers, but not all of our applications run that way,” he says. “The target is therefore the complete migration of all our applications that today run on a very large virtualised environment to a container-based environment. That goes hand-in-hand with the building out of a globally distributed persistence layer, where we have the ability to pick and choose where we produce our services, depending on customer requirements as to where they are consumed, and what security, privacy and compliance demands our customers have.

“The world is big, and the speed of light applies to Amadeus, too. Milliseconds used up over the wire to and from Australia can be vital.”

Mat Keep, director of product and market analysis at MongoDB, says Amadeus is using the database technology for more than a dozen applications, from back-office airline accounting platforms through to consumer-facing, mission-critical search applications.

“For many of its other MongoDB projects, Amadeus has also worked closely with our consulting services team,” says Keep. “In fact, the range of complex and industry-leading projects the company is working on has made it one of the most sophisticated users of non-relational technologies. Many best practices, and some features, have been created in concert with Amadeus staff.”

Read more on Business intelligence and analytics

Data Center
Data Management