ACL.
ACL.
ACL.
You might be using this pattern right now. Following it. Borrowing from it and you have no idea it has a name.
At NinjaCat, one of the first tasks handed to me when I joined, was making our MongoDB data warehouse more reliable and scalable. I did everything I could. Sharded replicated clusters. Jumbo data optimization. All the MongoDB greatest hits.
None of it mattered.
We were treating MongoDB like an RDBMS. We had roll-up queries that took minutes to complete.
So we migrated to Snowflake.
Here's where it gets interesting.
We had adapters everywhere. Hardcoded methods reaching into MongoDB from every direction. The frontend had methods declaring specific MongoDB keys in the payload. Years of accumulated integration debt, baked into every layer of the application.
The business requirement was simple: migrate the data, keep the interfaces and APIs untouched, adapt to new data structures.
Enter the anti-corruption layer.
The ACL sits between your new system and everything that still thinks it's talking to the old one. It translates. It protects. It lies, professionally, so your shiny new architecture doesn't get contaminated by legacy assumptions.
Your frontend still sends MongoDB-shaped requests. The ACL catches them, translates them to Snowflake queries, and returns responses shaped exactly like the old system would have.
The new system stays clean. The old integrations keep working. Nobody has to rewrite forty-seven microservices on the same sprint as a data platform migration.
ACL can be a small little microservice, a sidecar, even a business service layer.
The name explains the objective and purpose.
You're probably already using it.
❤️
Jake