The Challenge Of Scale

Nitish Khanapure, Vice President, Technology At Expedia Group | Friday, 16 August 2019, 13:12 IST

Nitish Khanapure, Vice President, Technology At Expedia GroupExpedia Group is the world's travel platform where we are leveraging our platform and technology capabilities on a huge scale across an extensive portfolio of businesses and brands to orchestrate the movement of people and the delivery of travel experiences on both a local and global basis. By way of example, we manage about 750 million searches a day from our travellers and lodging partners, and handle more than 9 billion pricing and lodging updates every day in multiple currencies.

This article describes the three areas that have had the greatest impact on our journey from the legacy stack to the existing state-of-the-art system that meets evolving business needs of our several product stacks.

Keeping Pace with the Traffic Growth

Migration to Cloud is a key part of our strategy to handle the internetscale traffic. Our Cloud migration is unlocking a new level of velocity. While we plan to operate our own data center footprint for core functionality for the foreseeable future, E we have initiated a companywide effort to migrate significant portions of our products and capabilities to the cloud computing environment over the next few years. We have aggressive plans for further migration.

“Cloud is not just about the elasticity or the buffet of services offered. Cloud can enable you to develop the DevOps culture”

Cloud is not just about the elasticity or the buffet of services offered. Cloud has enabled us to develop the DevOps culture. Our teams have become self-contained and autonomous, managing the development, deployment, monitoring and cost management for their stacks and services.

Modernizing the Platform

Not long ago, we were operating with a monolith architecture where teams had to work in lockstep for release cycles and changes were hard to push in. To solve this, we decided to move to the microservices architecture. With time, it became increasingly cumbersome to manoeuvre the structure and introduce changes alongside. To solve this, we started stepping towards a microservices architecture.

One of the big challenges was to evolve to this new architecture while the underlying software engine continues to power our business. This was akin to changing parts of a moving car. Today, microservices are the life force of the tech stacks at Expedia Group, and we are looking forward towards deprecating the monolith architecture. This has allowed us to create a federated release model for faster release of features to our customers.

We have also liberated ourselves from technology governance and shifted to a model where we define expectations and specifications around performance, monitoring, stability, cost, and similar operational metrics. The engineering teams perform the tech stack decision making with the overall goal to meet operational metrics. This is a big cultural shift from the monolith days that has helped us to scale our architectural choices and an important part of how we create a workplace and culture where exceptional people can do their best work.

Being Customer Centric using Machine Learning

The first 20 years of online travel put the power of booking a trip in the hands of travellers, but we suspect the near future of online travel will be about using technology to put more of the onus back into OTA (i.e., putting the Agency back into Online Travel Agency) for best traveller experience. Suggesting the best travel products to our customers is still a very complex matching problem we are striving to improve utilising various tools for online, Mobile, voice and chat iterations.

Machine Learning (ML) is redefining how we approach future developments. While Artificial Intelligence (AI) is already embedded within Expedia Group’s operations today we are only just scratching the surface of its applications and potential, as we move from machine learning towards machine intelligence.

We learned a great deal from data and trends. The combination of big data plus cognitive information and traveller insights about what helped them make a decision allowed us to make informed decision for our product development. By way of example, we have developed an image classification engine that uses deep learning neural nets to classify the hotel photographs provided by supplier-partners and travellers of the lobby, pool, room, etc with a high accuracy. The data-driven, test-andlearn culture has allowed us to learn, innovate, and sometimes fail or correct course.

It’s an exciting and humbling journey for our global team at Expedia Group as we look forward, stretch the limits and scale to support the growing business and global traveller needs.

Don't Miss ( 1-5 of 25 )