OLX Autos Consolidates Workloads on AWS to Improve Stability and Reduce Operational Complexity

The global used car market is projected to grow 8.3 percent from 2020 to 2027, with Asia-Pacific highlighted as a particularly fast-growing region. One of the reasons for this growth is the scarcity of semiconductor chips—critical in car manufacturing—due to supply chain interruptions in 2020. As a result, more people have started turning to used cars and online marketplaces to find their next vehicle. 

OLX Autos—part of the OLX Group—is a global marketplace for buying and selling used vehicles. The business aims to provide a safe, convenient one-stop solution for both buyers and sellers. Since launching in India in 2006, OLX Autos has expanded to 11 countries, including South America and Turkey. 

In 2019, OLX Autos merged with Frontier Car Group (FCG), a used auto marketplace in Germany. Until then, OLX Autos had been exclusively using Amazon Web Services (AWS) to host its cloud infrastructure. However, FCG was running on a different cloud provider. With the merger, OLX Autos became a multi-cloud company, which made operations more complex, costly, and time-consuming. Nikhil Sharma, platform engineering manager at OLX Autos, explains, “The biggest challenge with a multi-cloud strategy was an increased cognitive load on our engineers, because they had to build solutions compatible with both clouds.” 

Because workloads were distributed across two cloud platforms and four additional third-party software vendors, engineers were also not receiving enterprise-level support from any vendor except AWS. “This created chaos during production issues as support SLAs [service level agreements] from our other vendors were set at 24 hours, which was unacceptable. The business was facing serious challenges and we couldn’t do anything but wait for their support teams to respond,” Sharma adds.

OLX Autos sought to streamline its cloud infrastructure and in January 2022, the business completed migration of all workloads to AWS. The company uses AWS Enterprise Support and consistently receives a response to support tickets within 15 minutes, compared to the average response time on the other platform of 10–12 hours. 

Billing complexity is also reduced, which Sharma says generated a “sigh of relief” from the finance and accounting teams. “Because AWS is feature rich, we found everything we required in one cloud platform and no longer had to use third-party vendors for several core operations reliant on open-source protocols,” he explains. 

For example, OLX Autos uses Redis as an in-memory data store to improve application performance. With the all-in migration to AWS, the company could switch from a third-party Redis management tool to Amazon Elasticache for Redis. Likewise, it switched from third-party to managed AWS solutions for other open-source tools that are fundamental in the OLX Autos tech stack. These include Amazon OpenSearch Service to manage Amazon Elasticsearch and Amazon MQ to manage the RabbitMQ message broker.

In addition to leveraging native managed services on AWS, OLX Autos was able to optimize infrastructure costs during the migration. Engineers reassessed each workload before migrating, engaging in data-capacity planning to right-size the servers for each use case. The company forecasts annual savings of about $333,000 (€300,000) post-migration, in addition to $250,000 in credits in the first year after migration. 

Application and website performance was another persistent challenge in the multi-cloud environment. With the other cloud platform, OLX Autos users would report extremely slow response times on the website once or twice a week. The company traced the cause of these lags to a proxy server necessary for the other cloud platform to connect the primary OLX Autos application to corresponding databases. 

On AWS, however, database connections are direct, which eliminated the need for the proxy server. OLX Autos is using Amazon Aurora to manage the MySQL and PostgreSQL databases that power its website. Since migrating to AWS, the business hasn’t received any further reports from users regarding lags. 

Stability has also improved with the migration. A month before completely migrating to AWS, OLX Autos logged 2.5 hours of downtime. This has reduced to zero in the three months since migration, and engineers receive 75 percent fewer stability alerts. These alerts are triggered when users experience glitches on the OLX Autos platform or are unable to access the site. Zero downtime and fewer alerts further lighten engineers’ workloads. 

The cumulative effect of increased stability, integrated managed services, and faster issue resolution is an accelerated deployment and innovation cycle for OLX Autos. In recent months, the business has created an on-demand internal development platform using Amazon Elastic Kubernetes Services (Amazon EKS) that developers, operations, and product teams are accessing on a near-daily basis to experiment with new features.  

When OLX Autos was operating in a multi-cloud environment, it took around four to six weeks to launch in a new market such as Turkey. As engineers further refine the internal development platform, they are targeting a one-week time frame for a similar launch. “Our workloads in the other cloud were huge anchors weighing down our growth and inhibiting technology innovation. Since we’ve gone all-in on AWS, we’re focused more on our development platform and expect to reduce our time-to-market significantly,” Sharma says.

Sharma emphasizes the importance of having readily available support from a cloud partner as the business expands. He says, “Ultimately, stability supersedes everything. The completion of this migration brings a big smile to everyone’s faces, including top-level leadership, because we can confidently tell them our issues are resolved and support is one call away when we need it.” 

To learn more, visit aws.amazon.com/cloud-migration