Careem Scales Its Everything App Services to 50 Million+ Customers Using Amazon DynamoDB and AWS Database Technologies

Careem launched its ride-hailing app in July 2012 in Dubai. Since then, it has expanded into an everything - a single app capable of many different functions - that offers 12 daily services (including ordering food and groceries and sending money electronically) for customers in more than 80 cities. The ride-sharing service also grew. What started as a team of a few thousand drivers has grown to a company of a few million drivers. The app frequently records location data for the drivers, both while they’re driving and while they’re stationary. With this immense amount of data, Careem needed a solution that would scale with its number of drivers and customers while supporting the company’s growth into an everything app. “We’re getting locations from millions of drivers every few seconds,” says Khurram Naseem, senior director of engineering at Careem. “We needed a scalable solution for so much data, and Amazon DynamoDB was just about perfect.”

The company’s legacy database was built as a monolith. Careem realized that all this data funneling into one database was causing high latency and a lot of downtime when making changes requested by its internal teams. These uptime challenges could affect its customer base, which had grown from thousands of daily riders to millions by 2017. The company had been using AWS services since its inception in 2012, when it migrated from a monolithic database architecture to a microservices database architecture using Amazon database technologies. And to do so, it chose AWS Database Migration Service (AWS DMS), a managed migration and replication service used to migrate database and analytics workloads to AWS.

For its microservices, Careem uses several AWS services in addition to using Amazon DynamoDB. For example, Careem also uses Amazon Relational Database Service (Amazon RDS), a collection of managed services used to set up, operate, and scale databases in the cloud, for most of its workloads. The migration for this use case started with the ride-hailing service, but that was only the beginning of Careem’s journey using AWS. “We have more than 50 teams using AWS services,” says Naseem. “Today, we are virtually 100 percent on AWS and use more than 25 AWS services for analytics, data, and artificial intelligence across our app.”

The app collects data from Careem’s drivers, including location data, in Amazon DynamoDB. The company then uses algorithms built in house to find the driver who is best suited for a drive request. Careem worked alongside AWS teams to migrate from MySQL to Amazon DynamoDB and change its database infrastructure. “Different AWS teams were involved as we began to understand how to make improvements,” says Naseem. “We have changed our architecture and introduced roles to manage the volume of Amazon DynamoDB tables, which we have used to save on costs.” Careem has reduced costs for this use case to $1,600 per month from thousands, in part due to a schema-free approach to infrastructure.

Under the previous MySQL architecture, adding a single column or index involved downtime, which affected the entire service. Because Careem has to update its database with information on millions of drivers every few seconds, this downtime quickly added up. Amazon DynamoDB is more scalable in this regard, so Careem can add nearly any number of columns or fields in a particular row with virtually no downtime. “The way we are growing, there is a lot of scaling up because we need to track drivers by many attributes while they drive,” says Naseem. “By using Amazon DynamoDB, we were able to solve the problem of scalability.” One tracked attribute that is especially important in Careem’s operating regions is the ability to match female drivers with female passengers upon request. The company seamlessly manages this additional data, as well as the increasing numbers of drivers and customers, by using AWS.

By creating microservices, the company no longer has to worry about a single misconfiguration or deployment issue impacting the service as a whole. “We have used Amazon DynamoDB and the microservice-based architecture to reduce the number of issues compared with previous years,” says Naseem. This has improved the reliability of the ride-hailing app while Careem continues to expand its product and use AWS to grow it into an everything app. Careem also uses several other AWS services for its everything app, including Amazon Redshift, which offers optimal price performance for cloud data warehousing.

Teams at Careem have gained a better understanding of AWS services through the migration. Employees build their own perspectives by reading documentation and gain more insight by corresponding with experts from AWS, and then they use this knowledge to build up the database infrastructure. “When we start using a new service, we are very confident about talking to the AWS teams and explaining what we want to do,” says Naseem. “We have a lot of training available to help engineers grow and learn.”