GoDataFeed had a legacy platform built on on-premise MS SQL Server and tons of stored procedures. Scaling the platform required a lot of manual work and running concurrent tasks was still limited. Implementing new features, maintaining the codebase were also hard. Adding new customers to the platform was challenging because of the scalability issues and also was involving lots of manual work while bringing additional database instance license costs.
Based on their growth projection, their application has been modernised and rearchitected by leveraging AWS as Serverless. All stored procedures are re-implemented with .NET Core. The workflows were split into multiple tasks to manage them easily. On the database side, their relational data moved to Amazon Aurora Serverless MySQL and Amazon DocumentDB was used. Amazon S3 was used as the storage solution for keeping the intermediate files and AWS Step Functions were used for ETL needs which provide error handling and retry mechanism. For the business logic implementation, AWS Fargate and AWS ECS were used to run .NET Core applications without the overhead of managing the underlying infrastructure.
With the new architecture, the platform can scale up to hundreds of thousands of concurrent tasks; there are practically no limits for scaling. GodataFeed is now able to add new customers into the platform without any issues. The new implementation works on serverless, it's faster and cheaper compared to the legacy platform. Implementing new features is not painful anymore.
AWS Services:
Licensing Cost Savings: Windows Server, MS SQL Server.
Scaling: Infinite scaling with AWS Serverless Lambda.
HA(High Availability): Managed DB and cluster with HA capabilities.
Development Efficiency: Splitting the legacy to multiple modules.
Some of our results:
