skip to Main Content

Hotel Distributor Adopts Trio Data Engine for Enterprise-wide Analytics

This is the story of how our customer, a leading B2B Hotel Distributor selling 12 million room nights annually, adopted enterprise wide data analytics using Triometric, Elasticsearch and Kibana. Over the years, using its XML insights, this wholesaler became a forerunner in the B2B travel distribution ecosystem. This case study explores our customer’s challenges in incorporating XML data into their cross-departmental BI strategy and how our platform’s versatility led us to innovate a new approach to delivering analytics ready XML data.

This leading international accommodation wholesaler has been using the Trio Analytics platform for many years to enable the company to optimise its distribution operations by monitoring and analysing their XML API traffic. Travel distribution is a global and complex business with a network of suppliers, travel agents and wholesalers, each playing their specific role in the distribution of accommodation and tourism experiences. A network of APIs is the backbone for exchanging data such as inventory, rates and availability across the travel supply chain by travel search requests and replies that use the well-established XML protocol. Indeed hotel distributors today are largely technology companies delivering robust, high volume communication channels with an almost omnipresent reach.

Leading wholesalers, such as our client, know that the true power of distribution is in the ability to position the right products at the right time and price to the right customer. With so much information flowing through APIs, the Trio platform offers a central hub for operational and business insight that helps them achieve this.

Partnership – Where the Story Began

Initially the Hotel wholesaler used the Trio platform for managing IT performance related issues. By passively monitoring the API traffic the IT Operations team gained real-time dashboard visibility into all requests the company’s new API was getting and the health of their web services infrastructure to cope with those requests and corresponding replies. End to end response times, errors, anomalies, service degradation issues all clearly visible with the ability to drill down to root causes for swift correction.

Armed with this new intelligence, it wasn’t long before the wholesaler began using the Trio platform to analyse the content of the XML messages for business insights as well. By looking at indicators such as inventory levels in relation to demand or dates of travel, business analysts could better understand how far into the future web visitors were searching and for which destinations. Trio’s embedded reporting tool gave the analysts the visibility they needed to drive distribution decisions on a daily or even hourly basis.

Years of using the Trio API monitoring and analytics platform convinced the wholesaler that Trio is the best solution available for very high speed and high volume XML data capture, storage and preparation for analysis. So when the company embarked on a new initiative to develop an open-source enterprise wide business intelligence platform to serve multiple departments, it soon became clear that the Trio platform would remain the ideal data source for its very high volume API traffic. In the role of XML API technology partner, Triometric worked with the wholesaler to ensure that the XML request and reply message data was suitably processed, aggregated and made fit for piping downstream to the wholesaler’s chosen enterprise BI platform, the ELK stack.

ELK stands for (E)lasticsearch, (L)ogstash and (K)ibana which is an open-source technology stack designed for searching, capturing and presenting data from multiple sources and currently a popular choice among IT departments because of its ease of initial installation and versatility..

Elastic recommends that Elasticsearch, Logstash and Kibana are run on separate servers in a production environment so using the Trio platform for XML capture and preparation is part of a natural configuration. Trio also massages the log data as required by Elasticsearch to avoid mapping conflicts or field ambiguities so that data integration can take place with ease.

The Birth of Trio Data Engine

While Elasticsearch is clearly designed for searching large data stores, and Kibana for front-end visualisation, Logstash, the log file data collection component included in the stack proved not to be a good match for parsing and indexing the millions of API messages – the requests and responses are significantly more complex than the typical system and website log files normally processed by Logstash.

Travel requests and responses are complex in nature because they need to include a lot of detail. Destinations, dates, hotel codes and categories, room types, guest details and more are all part of the required information requests and the returning offers and choices. The Trio platform has been built from the outset, not only to provide operational insight but also to collect, parse, normalise and aggregate this type of data in real-time. Our XML data experts work with our customers to identify the key performance indicators that their business needs and making these readily available.

Since the wholesaler was already successfully using the Trio platform for XML API data collection and analysis, it was the logical next step to use the Trio platform for collecting and preparing XML or JSON data and feeding it into the company’s Elasticsearch instance, where it could form part of the search engine database.

Keen to meet our customer’s need for prepared XML data for use in business intelligence, the Trio customer team worked collaboratively with the wholesaler’s BI team to evolve the Trio data collection and analytics platform into an agnostic processing hub for transforming vast quantities of raw data into enriched and aggregated data fit for downstream analysis.

The resulting product was the first iteration of the Trio Data Engine that delivers the heavy lifting of collection and KPI extraction for XML data that can then be stored, searched and visualised by real-time BI platforms such as Elasticsearch and Kibana. It is a highly scalable and agile system that can evolve to handle continuously increasing high volumes of data.

An early decision by the wholesaler to use available on-premise hardware within the data centre quickly got overturned in favour of a dedicated cluster of hosted cloud infrastructure. This approach is the most flexible option for scaling the required resources needed for the growth in data and the inevitable rise in searches. XML API data in particular benefits from the ability to scale network and server capacity up and down at will. Cloud infrastructure is the best option for such elasticity, by making it relatively simple to add or remove the resources needed for data analytics capacity on demand.

Different analytics solutions often require different data formats or platform capabilities. Force fitting solutions to platforms can lead to high project costs and functional issues. Adopting the best platform for the right task helps to break down data silos and make relevant data more accessible. Using Trio Data Engine for high volume data collection and preparation, Elasticsearch for searching the data and Kibana for visualising the data provides a compelling platform that includes API analytics in an enterprise grade BI solution that scales.

The Vision – Single Source of Data Truth for All

trio data engine newsAs a global B2B wholesaler, this customer has always keenly invested in new technology and been conscious of the role that business intelligence can play in competitive success. Like many market leaders, the company wanted to move away from the limitations of data silos and towards a more enterprise wide approach to making data widely available to those that need it across departments. Instead of querying one data mart to get a few answers on inventory for example, the company wanted to be able to offer its cross-departmental teams the ability to go to one place for a much broader range of answers. This is a much more data centric approach to the business where efficiencies of scale kick in.

Our customer’s vision is to provide the same reliable trustworthy data to IT or business analysts whether they are focused on contracting inventory from suppliers, sharing rates and availability with distribution partners and customers, or simply just keeping the web services operation optimised.

By aligning data from multiple sources, and enabling multiple departments to access this data, executives can be sure that they are leveraging the most up-to-date details possible from sectors across the company. Putting custom dashboards in prominent office locations puts the key metrics for a team at the centre of their decision making. This is where the company starts to see the benefits of having access to real-time views, and making it available to everyone.

Benefitting from API Analytics

Like in many organisations, our client’s IT department is embracing a new way of delivering services to the business. Rolling out self-service analytics using the ELK stack and cloud-based applications is part of this disruption. Breaking down data silos and making data from different sources more accessible and connected is part of this journey.

XML-based API traffic is the lifeblood of this wholesaler’s online travel distribution business. The ability to track and match millions of API requests and replies in real-time is a key competitive differentiator. So it was important for this wholesaler to include API traffic and its detailed content analysis in an enterprise wide analytics view. For this reason, our client embarked on a project to use the output of the Trio platform as the XML data feed for the company’s cross departmental Elasticsearch project.

In summary, our Hotel Distributor client has deployed the Trio Data Engine to capture and translate raw API traffic data into aggregated and normalised data for piping downstream into an advanced analytics platform. This aggregated data source is then transformed into real-time and historical insights available to those that need it across operational and business departments. This analysis contributes to:

  • Data –driven decision making across distribution and revenue management
  • Increasing IT operational efficiency with pro-active monitoring
  • Ability to drill down to root cause for rapid problem resolution.
  • Ability to use data to track and react to market trends and opportunities

So whilst Elasticsearch is fast for searching data prepared as tables, and Kibana a popular and easy to use visualisation tool, the heavy lifting of XML capture and preparation is a pre-requisite necessary to get hotel distribution data in the right shape, ready to deliver enterprise value through effective BI analysis.

So, even if the agents connecting to our Hotel Distributor’s API remain unpredictable at least our Hotel Distributor is now well placed to tackle the ever changing business and market challenges that lie ahead.

Back To Top