CAFU Doubles Productivity by Modernizing Its Infrastructure Using Amazon EKS

Founded in 2018 in Dubai, CAFU is on a mission to save car owners time and trouble by providing on-demand mobile vehicle services. The company’s fleet of fuel delivery and car care vehicles can be seen all over the city, providing mobile refueling, tire rotation, and car wash services with the slogan “We bring the service station to you.” Recently, CAFU began gearing up to scale into new markets and verticals with expansions into Canada and Saudi Arabia.

Its legacy tech stack, however, was a major bottleneck. When engineers wanted to make a minor change to CAFU’s system, they had to consider the consequences for every aspect of the tech stack. Moreover, customers had to wait for up to 5 seconds for some endpoints to load. In short, the legacy architecture was hindering the solution’s availability and scaling—critical requirements for CAFU’s business.

CAFU set out to rearchitect its system from a monolithic stack to modern microservices with a yearlong initiative: CAFU 2.0. As the name suggests, this effort was comprehensive in scope and business critical in impact. What’s more, the startup needed services that could support a migration with near-zero downtime because it needs to serve customers 24/7 with acceptable latency.

CAFU was looking for a comprehensive service offering, including Internet of Things technology, and after evaluating a few options, it chose AWS for CAFU 2.0. “AWS has a comprehensive service offering that can support our business requirements,” says Yang. “It also has a great reputation, the adoption rate in the industry is very high, and it offers both scalability and advanced security features.” To automatically provision its new infrastructure, CAFU implemented Amazon Elastic Kubernetes Service (Amazon EKS), which lets organizations start, run, and scale Kubernetes without thinking about cluster management.

Using Amazon EKS, CAFU was able to adopt a wide range of services without complicating its deployment process. The company created a standardized development process that would free up its engineers to build the business logic. “Amazon EKS provided a great way to build and deploy a container image—then you’re done,” says Yang. “It was quite simple for us.” By standardizing its development process and using a fully managed Kubernetes service, CAFU significantly improved its time to market without overburdening its lean DevOps team.

The company uses many AWS services in its new architecture. For instance, it employs Amazon Relational Database Service (Amazon RDS)—easy-to-manage relational databases optimized for total cost of ownership—for each of its microservices. Using Amazon RDS, CAFU boosted the performance, security, and availability of its databases. The improvement in availability was especially significant—going as high as 99.99 percent and with virtually zero production downtime reported since the launch of CAFU 2.0. To enhance data processing and analysis, the startup uses Amazon Redshift, which delivers unmatched price performance at scale with SQL for data lakehouses.

By simplifying its infrastructure, CAFU reduced latency for some of its endpoints by over 50 percent, shortening wait times for customers. It also unlocked a new, sleeker UI for its app. In addition, it minimized downtime thanks to quicker bug resolution and reduced its infrastructure costs by more than 50 percent. “Our infrastructure spending is about 30 percent lower now,” says Yang. “We also don’t have to use third-party support solutions anymore, which saves us another 20–30 percent.”

Perhaps more importantly, CAFU significantly reduced its engineering costs per shipped feature. “We’re taking on more complex tasks and delivering them in a fraction of the time,” says Aadam Zaidi, head of product for the core platform at CAFU. Instead of dealing with infrastructure maintenance challenges, the team is focusing on change management so that each new delivery undergoes rigorous analysis and fine-tuning.

CAFU’s customers are now seeing rapid improvements to their experience. The company is targeting a 30 percent total reduction in service delivery time, and it is working to provide more accurate estimated times of arrival for its users. Moreover, engineers don’t have to worry about their infrastructure scaling to meet peak demand around rush hour. Instead, they can focus on adding new services to the customer experience.

With the CAFU 2.0 initiative complete, the startup is considering new ways to save its customers time and effort. CAFU has already experimented with artificial intelligence, and it is seeking to add vehicle telematics and Internet of Things connectivity to its operations. “The purpose of you owning your vehicle is to take you places, not to worry about where you need to take the vehicle, whether it’s a petrol station or car wash bay,” says Zaidi. “Our goal is to get better at taking that load off our users’ minds.”

Now that engineers can add and update services without worrying about infrastructure management, CAFU is ready to accelerate its innovation road map. “Using AWS, we can innovate, optimize, and continually improve our systems with confidence,” says Yang.