Application Migration
Migrate and Re-Architect Machine Learning Applications on Google Cloud Platform
A UK media services company with a focus on interactive planning and purchasing required a solution to migrate Machine Learning applications deployed on ShinyServer from Azure to Google Cloud Platform and re-architect the applications to reduce code redundancy, introduce scalability and enhance performance.
Client Challenges and Requirements
- Need for modern infrastructure to cater high availability and scalability requirements
- Code redundancy as same code base was coped for multiple clients which resulted in increased Go-to-Market time
- Certain operations were performance intensive resulting in increase page load time
- Manual deployment process
Bitwise Solution
Bitwise implementation resulted in migration of the application with the re-architected applications. Implementation included:
Data migration from SQLite and Azure VM local storage to Google Cloud Platform
Redesign of the Shiny application to have single version of the code bae to reduce maintenance overheads and increase Go-to-Market time
Role based access control
Build a platform for application testing – functional and performance and interactive debugging
DevOps Implementation
GCP infrastructure setup and optimization considering - security, scalability, performance and cost optimization
Tools & Technologies We Used
GCP Kubernetes / GKE
Azure ISTIO, NGINX
CloudSQL Docker
R, ShinyServer OAuth
Key Results
Kubernetes cluster setup for scalability and High Availability
Acceleration of Business Flows
DevOps Implementation
Foundation for Enterprise Platform
