LARGE FINTECH IN THE MIDDLE EAST BENEFITS FROM
AZURE TO AWS MIGRATION

Client Profile

Key Requirement

High-level Challenges:

• AWS EC2 – To run the web services consumed by the Web Application
• MYSQL database backup to be taken during non-business hours
• RDS to be used for MYSQL databases
• Databricks setup for ETL workloads
• ECR is to be used to store all the container-based images
• Application to be deployed behind the load balancer
• EKS cluster setup for the deployment of microservices
• API Gateway deployment to meet business requirements

Solution

Flentas provided a solution to Cassbana's need to migrate its Azure Cloud infrastructure to AWS Cloud quickly without downtime. Flentas proposed to engage with Cassbana to define the roadmap, plan, design, and execute the implementation plan, and work closely with them to deploy it on the AWS cloud.

• The proposed solution for setting up the Application on AWS includes the following services:
  • AWS EC2 – To run the web services consumed by the Web Application
  • Load-Balancer – To load balance on multiple EC2 Instances
  • RDS MYSQL – To Setup the MYSQL database on RDS with read replicas in AWS by migrating them from Azure
  • Amazon VPC – To segregate the network between multiple environments and application tier
  • ECR – To store container images
  • EKS Cluster – To run and deploy several micro-services applications
• Modernization of the Application to leverage object storage services by using fully managed AWS services like S3
• The migration of Cassbana's existing API and business logic was carried out with the help of API Gateway and Lambda Functions, allowing for increased efficiency and cost-effectiveness
• Migration of Data lakehouse using Databricks cluster setup in AWS for ETL workloads
• We set up Cassbana's Data Analytics platform using various AWS services such as DynamoDB, Amazon Quicksight, Athena, and Glue. It allowed for better data analysis and insights
• Planned the cutover for Cassbana applications and migrated their workload from Azure to AWS
• Establishing site-to-site connectivity with respective vendors
• Set up of basic monitoring system to alert the critical thresholds and Identify workload patterns

Components and Services

• AWS EC2
• Amazon Load Balancer
• DynamoDB
• ECR
• EKS
• Lambda
• Glue
• Amazon S3
• Quick Sight
• API Gateway
• Databricks (Marketplace)
• Application Load Balancer
• CloudTrail
• CloudWatch

Our Approach

• The decision was made to use the AWS Ireland Region for hosting all AWS infrastructure resources
• Each tier was further segregated into multiple subnets residing in each AZ available in the region for high availability setup
• All the applications that support Active-Active setup were placed behind a load balancer
• Azure Database dump has been taken and restored on the newly created RDS instance. Database users were created accordingly. A replica of the RDS instance was also created
• Site-to-Site VPN connectivity setup was done, which is used for Applications calling the partner APIs and databases
• Required docker images were also migrated from the Azure container registry to AWS ECR
• AWS CloudWatch was used for monitoring and setting up alerts and logs for all metrics of the infrastructure
• Databricks cluster setup was done from the AWS marketplace
• To set up an Amazon EKS cluster for microservices applications, created an EKS cluster and launched worker nodes into it. Then configured the worker nodes to join the cluster and deployed microservices applications using Kubernetes Helm charts
• Deployed a suite of AWS services to support data analytics use cases. It included AWS Glue, which is used for ETL (extract, transform, load) jobs to move and transform data into AWS data stores
• We also used Amazon Quicksight as our business intelligence tool for data visualization and reporting. We utilized Amazon S3 as our primary data lake to store and manage our data. We leveraged Amazon Athena for querying data stored in S3 using SQL queries

Cutover Approach

• API gateway was set up for all B2B applications, which allowed for forwarding the traffic to Azure endpoints
• After the Infra setup was completed, a DB pipeline was created to migrate the data from Azure to AWS
• After the App server setup got completed along with the necessary VPN connectivity, the Application was exposed via ALB
• In the Maintenance window, the Backend URL was changed to AWS in the API gateway to ensure no downtime was required
• For K8s services, cluster setup was completed first, and then one by one Application was deployed into the K8s cluster
• After deploying the Application, the respective application database was migrated to AWS
• Databricks cluster was setup from the marketplace, and Object storage data migrated to S3
• DB – DMS Pipeline created for migrating data from Azure to AWS
• EC2 – Application deployed manually on EC2, and data restored via S3
• K8s – Services deployed in cluster manually
• Jenkins - Pipelines were migrated separately, as it wasn't impacting migration

Measuring Success