Zoona Case Study

Zoona provides African entrepreneurs with the technology and resources to offer financial services—primarily electronic money transfers—to people in their local communities. To date, Zoona customers have transacted more than $1 billion dollars through these local entrepreneurs—known as Zoona Agents—who have earned combined commissions of more than $10 million. Zoona supports about 1,500 Zoona Agents, who have in turn created more than 2,500 jobs related to their financial services activities.

Launched in 2009, Zoona helps more than 1.6 million sub-Saharan Africans gain access to financial services in a region where, according to the International Monetary Fund, one in three adults is without access to these services. As Zoona’s business grew, the company identified several challenges with its IT infrastructure. The infrastructure was a hybrid solution with front-end applications in the cloud and its main databases running on bare-metal servers in a collocation site. A third-party service provider managed the database infrastructure and the Zoona IT team managed cloud-based applications.

A key challenge for Zoona was expanding the database platform with new hardware. It could take months to complete the procurement cycle and the deployment phase. What’s more, the platform had to be sized to handle the busiest period—in December, when Zoona Agents arrange the most electronic money transfers. In January and February, when the number of electronic money transfers were normally at their lowest, the hardware platform would be significantly underutilized.

It could also take weeks to complete basic management tasks, such as setting up virtual private network (VPN) connections to the databases for new Zoona Agents or making changes to the firewall. This was because of the time it took for multiple parties, including the third-party service provider for the databases, to coordinate their activities. Andre Penderis, chief technical officer of Zoona, says, “We wanted to increase our IT agility to make it easier to expand our operations. We also wanted to make our IT more scalable to reduce costs.”

Zoona also faced issues related to the performance and reliability of its IT infrastructure. Binoy George, technical operations lead at Zoona, explains, “The database platform suffered regular outages and we couldn’t get to the bottom of what the problems were. Network latency meant that Zoona Agents found their applications ran slow, with transaction speeds of about 1.5 seconds.”

Zoona planned to re-engineer its IT infrastructure and move to a microservices architecture. With microservices, applications are broken into their constituent components—for example, web access, business logic, and database. By adopting this approach, the Zoona IT team could build and maintain applications more easily.

The Zoona team planned to manage the company’s cloud resources using Terraform, a software solution that automates the provisioning of compute resources in cloud environments. It also planned to use Datomic, a distributed database that separates transaction processes, storage, and data queries. Pieter Breed, core platform lead, says. “When we looked at using Datomic, we found the tool was built specifically with Amazon DynamoDB in mind. All these factors—combined with my experience of working with Amazon Web Services—helped us decide to engage with AWS.”

The Zoona IT team migrated the entire IT stack supporting its operations to AWS. “The documentation available on AWS is very good and the AWS solutions architects we engaged with were very helpful and knowledgeable, says Breed. “Using that knowledge and our IT skills, we were able to develop our AWS infrastructure internally and migrate our main databases.”

The Zoona microservices architecture—including web access, business logic, and database—runs on Amazon Elastic Compute Cloud (Amazon EC2) instances. These instances support Zoona’s environments for development, user acceptance testing, quality assurance, training, and production. Zoona also uses Amazon Relational Database Service (Amazon RDS) to host its MySQL and PostgreSQL databases, which contain data on Zoona Agents and financial services transactions. The company runs Amazon DynamoDB as the backend for Datomic. Finally, Zoona uses AWS Identity and Access Management (IAM) to control access to the AWS services and resources by internal Zoona staff.

Today, Zoona’s IT is better able to support business growth and the company has eliminated its previous IT challenges. Zoona can scale up its database platform—or any part of its IT infrastructure on AWS—without concerns about time-consuming procurement cycles or deployment phases. Plus, the firm can scale down its IT to avoid underutilization during quieter months. “We can also scale down each evening when the amount of traffic significantly falls,” Penderis says. “That’s how flexible an AWS infrastructure is.”

Basic management tasks are also much faster. “Before, it could take us months to create VPNs for our Zoona Agents or to adjust the firewall, but with AWS, these operations take a matter of days,” says Penderis.

By moving to AWS, Zoona both cut its IT costs and increased IT agility. “We reduced our IT costs by 50 percent by moving to AWS, even though our IT infrastructure is much larger and supports a wider range of services than in the past, when we had our databases running at a collocation site,” Penderis says. “Our IT infrastructure is also more agile with AWS. We can pivot whichever way the business turns. So, if Zoona wants to enter a new line of business, we have the microservices architecture and the flexibility of AWS to build and deliver the applications needed to support that new line of business.”

Thanks to the high reliability of the AWS infrastructure, IT performance is no longer a concern for Zoona. Latency has decreased across the Zoona infrastructure, and Zoona Agents have seen an improvement in the responsiveness of their Zoona applications. Penderis says, “Agents are seeing better performance, with transactions through AWS taking no more than 200 milliseconds—about 87 percent faster than in the past with our hybrid solution.”