Chola is a leading financial services company that started as an equipment financing company and has since evolved into a full-service financial services provider that provides vehicle financing, home loans, home equity loans, SME loans, investment advisory services, stock broking, and a variety of other financial services to customers.  

The customer’s IT environment is primarily deployed on-premises and managed by a traditional IT management vendor. While the customer aspired to leverage the scale, reliability, and next generation of technologies available on the cloud, they needed a partner who could consult, create a roadmap, and implement solutions to meet their specific needs. 

The customer had co-hosted multiple applications in VMs at On-premises infrastructure and has tried to optimize by co-hosting different BU applications on VMs. During month-end operations, there were frequent scaling issues, resulting in downtime that impacted multiple BU applications leading to business disruption.  

Any fixes or rollouts also had similar risks, which, when materialized, frequently impacted the business. Rollouts had more frequent deployment issues due to different testing environments that did not match the production deployment architecture of multiple applications co-hosted on VMs. 

The business application team wanted to migrate to the cloud to resolve these challenges. However, a business case could not be built based on a lift and shift migration approach as it proved cost-prohibitive. So, the customer needed to modernize and adopt a scalable, server-less, cost-effective solution that is also a steppingstone in their mid-term goal of upgrading their applications to micro-services architecture while minimizing application refactoring efforts in this phase while solving the problem at hand. 

Through a series of workshops, 1CloudHub understood the customer’s current application development lifecycle, the team structure, their skillsets, and the deployment model employed. 1CloudHub derived a target state deployment architecture, an updated DevOps tool set, and processes based on these factors.   

Aligning with the end goal of a micro-services architecture while addressing the current scale and reliability issues, EKS on Fargate was identified as the choice of platform for the application servers. At the same time, the databases were modernized to use a PaaS first approach with RDBMS landing in RDS and MongoDB in Managed MongoDB on AWS. 

Similarly, to deploy their applications to EKS Fargate, the DevOps tooling had to be updated to include CodeCommit as the git repository and ECR as the container registry. The existing Jenkins pipeline was updated to orchestrate deployment using these in Non-Production and Production environments. 

From a day two operations perspective, the monitoring instrumentation had to be updated to use Prometheus and Grafana. For container log management, CloudWatch Logs was employed. 

Finally, the application teams were educated about the need to use stateless services when deploying to EKS Fargate and helped them handle issues around session and application or runtime log storage (which until then was stored on the local host). Once the application teams (including the vendor teams) had made these changes, we helped them deploy to the new environment, test it, and then roll it out to production environments with business approval. 

• Elasticity to meet business demand during month-end processing: Scaling issues were fixed thanks to the Horizontal Pod Scaler (based on CPU utilization) in EKS Fargate. The month-end process is scaling without a hitch. 
• Improved system reliability, from regular firefighting to a focus on innovation: Moving away from a monolithic and co-hosted application deployment model meant that an outage on one server or component did not affect multiple business systems. The architecture also improves testing quality since the production, and non-production environments are now much liked. The architecture is promised to deliver an uptime SLA of 99.9% for the overall system, giving the application team more time to focus on improving business features and move towards micro-services. 

• Cost Efficiency: Modernizing to serverless and PaaS platforms on AWS brought down the infrastructure cost by 15%. 
• Operational Efficiency: The use of cloud-native platforms like EKS, RDS, and other managed services means that the operational tasks are automated through the System Manager. 

• EKS-Fargate
• Jenkins
• WAF
• S3
• RDS
• ECR
• CodeCommit
• AutoScaling