Atomstate Implements Customized Database DevOps for one of the largest Indian MNC private lenders.

The Client

The client is one of the largest Indian private lenders with operations in around 17 countries.

The Challenge

The client’s wealth management unit was looking for an automated solution to deploy database changes to multiple environments using a devops pipeline. The platform which we dealt with is a CMS platform whose backend is based on Microsoft SQL Server.

Their biggest issue was in UAT as this environment alone gets 5-6 changes per day. It had a complexity where in one change request may or may not be dependent on other change requests. A change request dependency chain could be a tree spanning 100s of branches.

Their previous approach was manual in nature. The wealth management team used to verify the deployment statuses of every dependent change requests. If there are 100 dependent change requests, then all 100 change requests would be verified. The latest change request would be deployed only if dependencies are deployed. Otherwise, it was an aborted deployment.

It used to take an average of 40 minutes per UAT deployment. Rollback was a separate pain altogether.

The Solution

Atomstate has delivered a built-from scratch automation CLI that has eliminated every bottleneck of previous manual approaches.

Our solution has specifically delivered the following

• Customized CLI tool - We have architected and built this automation CLI from scratch. This CLI has delivered end-to-end automation removing every manual interaction along the way.
• Delivery via Pipeline - We have set up a pipeline job in Jenkins, and it invokes the CLI to deliver the goods. Now there is no more unnecessary windows server login. Thereby reducing risks.
• Robust Logging - The CLI streams logs to elasticsearch datastore to keep the audit posture healthy.
• Dashboarding - The other benefit of elasticsearch integration is dashboard view in Kibana of every deployment for important stakeholders. The team is now able to take data driven decisions to plan better.

The Impact

• The deployment times have reduced from ~40 minutes to 2 minutes on an average. The team is now dedicating their time to pressing tasks and issues.
• The CLI automatically scans through deployment statuses of dependent change requests and decides on deployment of new change request. It has eliminated the dreaded and laborious manual validation checks.
• Now the system can handle any number of deployments in a day.
• The CLI does the automated rollback and adjusts the dependent maps by itself.
• A proper process has been instituted under this approach.