How does one of the world’s largest travel technology companies migrate to Amazon Web Services? Before you can even think about building your cloud infrastructure, there are dozens of critical design choices to be made. We supported one of Expedia’s successful travel technology brands to make the transition.
Expedia Partner Solutions (EPS) is the Expedia Group’s B2B partnership brand. DevOps principles have enabled them to remain market leaders in the industry, with around 600 code releases a year when they first contacted us.
Until a few years ago, the sheer volume of daily traffic had made on-premise hosting the most cost-effective option. Keen to benefit from the agility of a public cloud, they decided to migrate their successful operations into AWS. The new infrastructure would need not only to be PCI DSS compliant, but also comply with SOC2 requirements.
Although the team had deep operations and DevOps engineering experience, they didn’t have the depth of knowledge about cloud infrastructure that they felt they needed.
The Expedia Group had already agreed that AWS would be the best option for them and their brands when they moved into the public cloud.
Why The Scale Factory?
EPS contacted us initially to help them evaluate their options and make decisions about the design of their infrastructure. They’d done extensive research into cloud infrastructure, but needed deep experience to help them understand the implications of the choices they had to make. More specifically, they were unable to agree on the best choice of container orchestration platform and the tooling they would use to scale-up the migration.
One of our principal consultants worked closely with the operations team, answering questions and making design suggestions. They also ran a number of workshops with EPS developers to explore and explain the different design options and how they could make them work.
AWS ECS was chosen as the container orchestration platform, used with AWS EC2, while AWS CloudFormation was chosen to simplify infrastructure management and facilitate self-service provisioning, and AWS Lambda for increased responsiveness without the extra operations overheads. For encryption and authentication we used Hashicorp Vault.
An element of the project was to design the infrastructure for migrating the company’s existing data warehouse. For the new infrastructure we used AWS EMR with Qubole and Apache Kafka for building the real-time streaming data pipelines critical to EPS’s core business.
Once the design decisions had been made, we embedded two of our site reliability engineers within two different EPS teams to help them build their new cloud infrastructure.
The EPS team were able to make all the critical cloud infrastructure design decisions and had their first production system running on the cloud within 3 months of our initial engagement. They had migrated over 70% of their applications to the cloud within 2 years of starting to work with us, which considering the scale of their operations, is a huge achievement.
This was made possible by their infrastructure being designed to support automation and DevOps best practice, making it possible for their development teams to self serve without compromising security or efficiency.