Fast-growing Qatari tech company Snoonu migrated and consolidated key workloads and systems from multiple cloud providers to Amazon Web Services (AWS) to simplify staff operations, improve reliability, and save costs. The migration also gave the company the chance to re-engineer many workloads away from virtual machines and onto Kubernetes on AWS to take advantage of microservices architecture, reducing costs and minimizing management overhead. Snoonu is also using AWS machine learning services to automatically improve its inventory management and customer experience, and to reduce manual overheads.

Snoonu Simplifies its Cloud Infrastructure for Reliability and Innovation

Snoonu is a fast-growing Qatari tech company with a 30 percent share of the domestic market. Through its app it offers customers deliveries from restaurants, its own market and pharmaceutical services, and various other grocery, fragrance, and electronics providers.

The company had various workloads running with different large-scale cloud providers but was experiencing performance and cost control challenges. In particular, its FalconFlex application—which runs driver dispatch and package matching—was running too slowly and suffering availability issues. FalconFlex is central to the smooth running of the business and needs the highest levels of service availability.

Snoonu realized its FalconFlex application needed re-engineering and decided it would be a good time to also migrate and consolidate workloads with a single, different cloud supplier. It chose AWS because it offered the scalability it needed as a startup business with uncertain growth prospects. AWS also had the services it needed to support all aspects of the business and its plans for the future. Once the migration was complete, Snoonu also wanted to use AWS machine learning (ML) services to automate inventory management and to improve customer experience by delivering more relevant product suggestions.

Migrating with Minimal Downtime

Snoonu chose to work with AWS Partner Zero&One to help with the migration. “They had great technical knowledge and helped us find the right assistance when we needed it,” says Nikita Gordeev, chief technology officer at Snoonu. “They also helped us take advantage of the AWS Migration Acceleration Program, which provided help with planning and discounts with the actual migration.”

Snoonu had three main infrastructure goals for the migration to AWS. First, to standardize on one cloud platform to make life easier for staff and help them build relevant skills. Second, to optimize costs and ensure Snoonu made best use of scale discounts. And finally, to build an architecture that could scale easily using Kubernetes and would be easily replicable to support Snoonu's international expansion. Kubernetes on AWS allows Snoonu to deploy and manage containerized applications and scale them easily. It manages Snoonu's Amazon Elastic Compute Cloud (Amazon EC2) compute instances and runs containers on them.

Gordeev says he appreciated the guidance and support Snoonu received from both AWS and Zero&One. In total, it migrated 300 GB of data—including a highly-specialized, self-hosted Mongo DB database—more quickly than it expected and with no unplanned downtime.

Better Performance for Lower Costs with Greater Visibility

The business-critical FalconFlex application now runs on Amazon EC2, which provides secure, resizable compute capacity in the cloud. The scalable systems have so far delivered 99.99 percent service availability.

With the migration complete, Snoonu worked with Zero&One to optimize costs. Together the companies reviewed the utilization of resources and made changes to instances to improve efficiency. Switching to newer and more efficient instance types is saving the company $10,000 a month and the cost-saving process is ongoing.

Beyond simple savings, the FinOps project with Zero&One has also helped the team gain a better understanding of its technology footprint. “It's given us a better understanding of our infrastructure and how to manage it and keep making cost savings going forward,” says Gordeev.

Using Machine Learning for Easier Inventory Management

Snoonu also chose to work with Zero&One and AWS because it wanted to use AWS ML and artificial intelligence (AI) services to make it easier for merchants to add new items to their inventory. The team developed a system that uses generative AI models to check new inventory entries.

It makes sure the image and product description match, using Amazon Rekognition for image analysis, Amazon Comprehend for natural language processing, and Amazon Bedrock for automatic categorization. These checks previously took up a lot of staff time manually confirming entry accuracy and making changes when merchants updated catalogues.

The solution will also provide automated content moderation, ensuring a safe user environment by using ML for accurate content categorization. AWS flexibility and scalability will help meet Snoonu's changing business needs as the company grows.

Better Security Posture, Improved User Access Controls

Zero&One also improved Snoonu's single-sign on authentication with a better set up of role-based security and access permissions. The company uses AWS Control Tower to set up an AWS Well Architected, multi-account environment and automate the creation of AWS accounts with built-in governance.

Using AWS Control Tower, security and access tasks that used to need help from DevOps engineers are now available to the IT team with the click of a mouse. Amazon Inspector also protects Snoonu's AWS workloads by providing automated and continual vulnerability management at scale.

Since the migration and re-engineering work, the performance of key applications has improved and the company has not experienced any availability issues. Snoonu now has systems that can continue to be improved for further cost savings and it is able to quickly spin up services in new countries and AWS Regions as it begins its international expansion. “We're confident now in our systems and our ability to manage them,” says Gordeev. “And we know that we can spin them up in any AWS data center in the world with no nasty surprises.”