Success Story
Producteca migrates its last critical services from Azure to AWS in 7 months, without refactoring production code
At a glance
Producteca is an Argentine SaaS platform that centralizes e-commerce operations management. Its last three critical services were still on Azure, with a search engine and messaging that were incompatible with their AWS equivalents. Craftech migrated them with an abstraction strategy that avoided touching the existing code. The result: gateway, messaging and search on AWS, without refactoring production code.
3
critical services migrated from Azure to AWS
Zero
production code refactoring
Producteca is an Argentine SaaS platform that centralizes e-commerce operations management: it integrates multiple sales channels and lets teams manage catalogs, orders and listings from a single hub. It operates mainly in Latin America.
Solution implemented
- Initial assessment of Azure and AWS
- Azure AI Search to Amazon OpenSearch Service: a Node.js wrapper that translates OData into DSL, deployed as an API in the same Kubernetes cluster, and index mapping with support from AWS specialists
- Azure Service Bus to Amazon SNS and Amazon SQS: automated extraction of topics and subscriptions, AI-assisted filter translation, and Terraform for every resource
- Application Gateway to an AWS Application Load Balancer routing to multiple Amazon EKS clusters
- Team documentation and training
Business Challenge
Producteca's last three critical services were still on Azure. The search engine was incompatible: Azure AI Search used the OData protocol, which does not exist in Amazon OpenSearch, and adapting the microservices that consumed it meant a massive refactoring. In addition, Azure Service Bus handled dozens of topics with filters incompatible with Amazon SNS/SQS. Everything had to be migrated without disrupting the business.
Solution
After an initial assessment of Azure and AWS, Craftech applied an abstraction strategy that avoided touching the existing code. For search, it built a Node.js wrapper that automatically translates OData queries into the Amazon OpenSearch Service DSL, deployed as an API in the same Kubernetes cluster, so the microservices kept working unchanged; index mapping was done with support from AWS specialists. For messaging, it automated the extraction of Azure Service Bus topics and subscriptions, translated the filters with AI assistance, and used Terraform to generate hundreds of Amazon SNS and Amazon SQS resources. Traffic balancing moved from Application Gateway to an Application Load Balancer routing to multiple Amazon EKS clusters. The project closed with team documentation and training.
Outcomes
- All 3 critical services (gateway, messaging and search) migrated from Azure to AWS
- Zero production code refactoring
- Hundreds of Amazon SNS and Amazon SQS resources generated automatically with Terraform
- Team enabled to operate on AWS autonomously
- Deliverables validated with AWS specialists
Stack & technologies
