Migration To GCP And Implementing DevOps Practices for SAVY

To make migration to GCP successful, a staged migration approach was chosen with application refactoring in order to containerize and apply DevOps practices. Cloud Monitoring and Logging were used to increase the quality of observability for the whole stack.

The challenge

After a few years at a traditional IAAS platform provider, SAVY saw that further growth was hindered by the limitations of the current provider, where dynamic scaling was not available, and any reconfiguration caused multiple-hour downtimes. The goal was to move to a platform that can scale on demand, be configurable by code, support containers, and offer managed services for databases and data warehousing.

The solution

At the first stage, the database was moved to CloudSQL, which opened the possibility to start developing a BI platform based on BigQuery and Airflow orchestration hosted on GKE. At the second stage, web applications were containerized and deployed on Cloud Run using GitLab pipelines, and Cloud Scheduler was chosen to initiate all cron jobs. With the last stage, static assets and recordings were moved to storage.

The results

After migration to GCP and implementing DevOps practices, SAVY can now develop new features and products without unnecessary holdups and downtimes from the infrastructure side. With the ability to have multiple development and verification environments, new feature releases become more frequent and of higher quality. Security and reliability were increased using a global load balancer and Cloud Armor.

---

SAVY is the first peer-to-peer lending platform in Lithuania, enabling people to lend and borrow money directly from each other. The platform is supervised by the Bank of Lithuania and follows Responsible lending regulations. Responsible borrowing and lending are inseparable from economic growth, and it helps to create greater financial freedom.