OLX Autos Delivers a Seamless Online Marketplace by Running Containers on AWS

OLX Autos has a mission to empower people to upgrade their lives through its online marketplace of services and used products that deliver superior customer value. As part of the OLX Group, which operates more than 20 brands across 30 countries, OLX Autos combines the agility of a startup with the maturity of a global entity. With a mission to shape the future of trade, OLX Autos is passionate about the use of cutting-edge technology to enhance the customer experience.

India is among the top growth markets for OLX Autos globally. The company has a highly innovative technology team running its Panamera classifieds marketplace, which is active in 12 countries, on the Amazon Web Services (AWS) Cloud. To unlock maximum value for its users, OLX Autos is committed to investing in its people and the latest technology.

In 2018, OLX Autos experienced an outage on its site when its internal certificates for OpenShift containerization software expired. The business was using OpenShift version 3.5 to manage its infrastructure and to deploy applications on its site running in Docker containers. As part of the company’s commitment to support customers with a smooth experience, engineers quickly identified the issue and rolled out new certificates immediately. However, they had to invest a significant amount of time to restore the website. The team concluded that their staging environment was unstable due to issues with the control panel on OpenShift and began seeking ways to modernize their technology stack.

Abhishek Tomar, infrastructure head at OLX Autos, says, “Buyers and sellers could not search, see new ads, or create new listings during the outage. We realized that even if we experienced just one crash a year, it would still be better to offload the control and management plane to AWS. Our customers were not able to access our marketplace in the several hours that were spent resolving the issue, resulting in customer dissatisfaction.”

OLX Autos engineers needed to rapidly upgrade the site’s architecture and decided to perform a proof of concept for hosting a self-managed Kubernetes cluster versus using Amazon Elastic Kubernetes Service (Amazon EKS). “We decided on Amazon EKS because it was more flexible than the OpenShift cluster,” says Nikhil Sharma, the chapter lead site reliability engineer (SRE) at OLX Autos who led the migration.

OLX Autos initiated a discussion with the AWS team a year before beginning the migration. “First, we reached out to AWS Enterprise Support for advice on the best migration strategy,” says Sharma. OLX Autos was simultaneously upgrading its operating system, Kubernetes, and Docker versions, and the complexity of its OpenShift architecture in addition to the software version limitations made the migration particularly challenging. The OLX Autos SRE team planned the migration strategy and started moving their staging environment to AWS in January 2020. “We got immediate support from AWS to resolve any issues in the staging environment, which helped us migrate smoothly to a production environment,” Sharma says.

The AWS team presented a roadmap for potential new Panamera features using Amazon EKS in addition to other AWS services. OLX Autos then started building an Amazon EKS cluster and migrating selected workloads from OpenShift, with the rollout completed in just a few weeks. OLX Autos has now migrated all 124 of its microservices to Amazon EKS, which has improved efficiency and scalability while unlocking cost savings.

Since the migration, OLX Autos infrastructure teams are now better placed to meet key internal product development deadlines. The business was initially using the Puppet configuration tool to manage its OpenShift software. Engineers needed to rely on their Puppet skill sets and spent three to four days a month rolling out and monitoring important updates to the OLX Autos infrastructure.

With the AWS migration, the company has removed Puppet from its architecture and offloaded container management to AWS. By running on Amazon EKS, the OLX Autos website benefits from improved performance and scalability, and engineers can reallocate their time to higher value-added tasks.

When OLX Autos was running an old version of OpenShift, the company experienced issues with autoscaling, which caused interruptions to its applications. “Now with Amazon EKS, our Amazon Elastic Compute Cloud (Amazon EC2) Fleet gets autoscaled in a couple of minutes whenever any marketing campaigns cause a surge or a sudden spike in traffic. If online traffic drops, the fleet gets scaled down—something we were not able to do smoothly before with OpenShift,” says Tomar.

OLX Autos has also offloaded the critical task of administering Secure Sockets Layer (SSL)/Transport Layer Security (TLS) certificates to AWS Certificate Manager. Previously, teams had to manually procure and deploy new certificates each year, but with AWS, new certificates are deployed with a few simple API calls. “It’s a relief that AWS will now be taking care of that,” says Tomar. “We don’t have to invest a single minute in checking and updating certificates, which could impact key timelines for our business.”

OLX Autos had booked a number of Amazon EC2 Reserved Instances for type C5 instances but was unable to use them with the older version of OpenShift. “Because of this migration and the version upgrade, we migrated all our C4 instances to C5 and could leverage Reserved Instances pricing. That has ultimately saved us 33 percent of compute costs,” says Tomar.

OLX Autos has also benefited from the AWS container network interface (CNI) plug-in, which was blocked with its previous OpenShift configuration. The CNI plug-in has not only improved application latency, but OLX Autos also expects to save at least 10 percent on its overall monthly AWS bill with more efficient applications. Tomar says, “Using AWS has kept our environment running in a stable manner in the most cost-optimized way.”